Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
How to rip from vinyl or tape

By mcgrew in Technology
Wed Sep 15, 2004 at 10:28:34 AM EST
Tags: Music (all tags)
Music

So, you have five hundred tapes and albums of the Beatles, Led Zeppelin, and the like, and want then on your computer? Well, you could spend hundreds of dollars replacing music you already paid for once with CDs. Or, instead of paying twenty bucks per album, you could spend twenty cents per album to convert your tapes and vinyl to CD and rip those. Read more for how.

Also, the instructions below will allow you to defeat any copy protection. Period. Just substitute "cheap CD walkman" for "stereo." Caution- doing this in the United Statesor other parts of the world may be a felony. Those in less corrupt, more civilized parts of the world need not fear.

Or alternately, you can make CDs of your own band.


Now, in many if not most cases you will not quite have the quality of CD you would have had if you had simply spent the money on brand new CDs. However, in some cases you can actually have a better quality CD than you can buy! This is because in a few cases, excellent audio engineers took great pains to make the LP sound as good as it possibly could. When remastered, these exceptional quality albums must have the treble tones attenuated to minimize the aliasing distortion. This causes the bass to be too loud, so that, too, must be attenuated. CDs simply do not have the undistorted frequency response of a quality cassette recorded on a good tape deck, which can record frequencies up to 18 khz. Vinyl is even better, with a frequency response high enough that they modulated music with a 44KHZ tone to make "quadrophonic" four channel stereo in the 1970s.

Led Zepplin's "Presence" lacks presence on the CD. Sample your own and you get the presence back- along with a noticable alias. But it still sounds better than the factory model.

To sample cassettes, you need only a cassette deck. If you don't have a good one, do not buy a new deck! Newer cassette decks sound terrible, some with frequency responses that don't go up past 3khz or down below 500hz. Even though its head will likely be a little worn, a used deck from the height of analog (1970-1980) will still sound better than the cheap junk they sell these days.

To sample vinyl, you will need a turntable and a preamp. The reason is if you plug a quality turntable directly into your sound card, you will have a very low volume, very tinny signal. This is because of a technology known as the "RIAA rolloff," which was a kind of "dolby" for turntables used long before Dolby patented the process. Bass tones need to be amplified, and the treble tones need to be attenuated. A good stereo reciever from the 1970s or 1980s will accomplish this for you.

If you have no equipment, you can usually find it in pawn shops or used record stores. It is seldom very expensive. I bought a used German made Dual turntable (with its tone arm supported by a 4 point gimbal, like a ship's compass) in town for fifty dollars. This same model sold for several hundred dollars when new.

You will, of course, need a computer with a sound card with an auxillary input. The make of the computer and its operating system do not matter, although for our example we will use a Windows computer. You will also need software to sample the music with. I use a program called Exact Audio Copy, a free program you can download from the above link. Although, as I said, you can use any computer, OS, or sampling program, the step by step instructions presented here apply to EAC on Windows.

Now that you have the hardware and software, it's time to connect the hardware. There are a few different ways you can do it.

You will need a stereo cable with RCA plugs (or a din plug if that's what your stereo uses) on one end, and a 1/8 inch stereo plug on the other end for plugging into your sound card. This often comes with your sound card, but if not, you can buy one at any electroniucs store (Radio Shack, Circut City, Best Buy, etc).

The 1/8th inch stereo plug goes into your computer's AUX IN jack. If your stereo has tape inputs and outputs, connect the other end to its tape output (often marked "record"). If you do not have a tape output, you can use your stereo's AUX OUT plug. If it has neither, you can buy an adaptor and use its headphone jack. Only use the headphone jack as a last resort!

Now that your hardware is all set up, start your software. From here on in, we aill assume you are using EAC (it's free), although again, any software should do.

First, start the volume control program. In it (example here is Windows) select "Options," then "Properties." make sure the correct input is checked for display, and under "Adjust volume for" select "Recording" and click "OK". On the "recording" control, "Mute" is replaced with "select". Make sure the correct input is selected (Line In or Aux, depending on which one you plugged your stereo into).

Start EAC, and under its "Tools" menu, select "Record WAV". A dialog with buttons and a level meter will come up. Click "Select Target File Name" (everything else is disabled). Give your recording a name (the default is "Record.wav") and choose where the file will be stored.

Before you actually start recording, you will probably want two records from whoever you are sampling, as you can usually fit two LPs on a single CD.

Now, find the loudest passage on your album and adjust the volume to get the meter as close to zero as possible without going over zero. If you go over zero, you will introduce very, very ugly "clipping" distortion, just like recording to tape.

Some people put the volume well below zero. This is nearly as much of a mistake as letting it go over zero, as the lower the volume, the more aliasing. If you record your album at half volume, you have essentially recorded it in eight bits rather than sixteen. You want it as close as you can get to zero without going over.

Now that the volume is set, rewind the tape (if you are using tape). Then, click the "Start Record" button on EAC, and start your album. Unlike a tape recorder, if you mess up the beginning, you don't have to start the recording over- simply start the record over. You will need to delete some silent sections later anyway.

When the first side is done, turn your album over and play the other side. Don't stop the recording! When this is done, start the second album, if you are putting two albums on one CD. When you are done recording, click "Stop Record" and then "OK" in EAC's recording dialog.

Next, from EAC's "Tools" menu, select "Process WAV". Select the file you just finished recording.

Now, highlight any blank parts and delete them (Edit, Delete Selection).

Then save the file, and select "Cue Sheet" from the menu, then "Create Cue Sheet".

You could select "generate cue sheet," but often this will miss some track starts. My preferance is to select track starts manually. Zoom into where a song ends (there will usually be a gap in the waveform, although not always). Place your cursor between the two songs on the waveform display, and click "Cue Sheet," "insert," then "track start" for the beginning of each song.

Then save your wave file (File, Save) and your cue sheet (Cue Sheet, Save Cue Sheet). Close the wave editor, and from EAC's menu select "Tools," "Write CDR." A dialog to write a CD will open.

Select "File," then "Load Cue Sheet." select the cue sheet you just saved.

From here, you can add more .wav files as tracks if you wish, or just select "CDR", "Write CD". If your music pretty much fills the CD, tell it to close the CD when done. If you're only putting a single album on it, leave the CD open and you can add a multisession volume to it with MP3s, HTML, pictures, or whatever. However, EAC only does audio- you will have to use whatever software came with your burner to add the computer-only files.

When it has finished, you will have a more or less CD quality CD, depending on your source. But we're not done yet- we still don't have any MP3s.

Now, for ripping MP3s I like a different program, CDex, although we'll continue with EAC for now. Open a web browser, go to Google, and search for "'band name' discography". When you find a track listing for your album, copy and paste into the proper fields in EAC or CDex (or whatever your favorite ripping program is). To change "Track 1" to the proper song title, just click on it and it will allow you to paste the real song name in.

Highlight all the songs, then click the "MP3" button. Your CD will then produce an MP3 of each song, with properly formatted ID3 tags.

Your MP3s will have every bit of quality as any MP3 of the same bitrate that was ripped from a twenty dollar factory CD. Your CD may well have even better quality than that factory CD! You've saved a little money, and had a little geeky fun in the process, as well.

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Poll
Most of my music is on...
o Vinyl 6%
o Cassette 1%
o Eight Track 0%
o Reel to Reel 0%
o CD 20%
o MP3/Ogg/WiMP 67%
o Wax 1%
o I only listen to live music 1%

Votes: 59
Results | Other Polls

Related Links
o Google
o Exact Audio Copy
o CDex
o Also by mcgrew


Display: Sort:
How to rip from vinyl or tape | 140 comments (112 topical, 28 editorial, 3 hidden)
Alternative (3.00 / 10) (#2)
by jmzero on Mon Sep 13, 2004 at 12:40:37 PM EST

If you're ripping from cassette tape (eeeuggh!), you might want to use specialized tape ripping software like Kazaa or Bittorrent.  There is some legal disputes around these softwares, as they can apparently also function without the cassette.  

However, it seems to me that being in possession of the LP or cassette might assuage whatever ethical (if not legal) concerns you may have about using said software.

As to LP, I recommend going along with the author - although you're going to want a very good setup and quality components if you intend to get a workable result.
.
"Let's not stir that bag of worms." - my lovely wife

If your cassette deck sucks (none / 0) (#31)
by mcgrew on Mon Sep 13, 2004 at 07:51:45 PM EST

then downloading the mp3 would, in some circumstances, sound better. Note the part about newer decks vs older ones as well. A quality tape recorded on a qualit dect using appropriate dolby settings (rare in a commercial tape, I know) sounds damned good, much better than even a 192 kbps MP3.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

That's just wrong (none / 1) (#35)
by CaptainSuperBoy on Mon Sep 13, 2004 at 09:49:33 PM EST

First off, I guarantee you can't tell a properly encoded 192kbps MP3 from a CD.

Second, there's no way you can approach CD quality by playing a tape into a cheap sound card. Even if you have a good tape deck and a good ADC, analog is never preferable. Cassettes served their purpose, but let's not get nostalgic about it: CDs sound better, period.

Also, the only appropriate dolby setting to record on is "none." DNR was introduced to get rid of hiss on cheap decks, but it also gets rid of signal. If you have a good deck you don't need it.

--
jimmysquid.com - I take pictures.
[ Parent ]

The Nirvana CD... (none / 0) (#89)
by mcgrew on Wed Sep 15, 2004 at 06:54:35 PM EST

sounds better than the Nirvana cassette, true. That's because the master was digital, so the cassette has all the disadvantages of both digital and analog, and the advantages of neither.

Likewise, the vinyl of Magical Mystery Tour (assuming your copy has only beeen out of the package a few times) will sound much better than the CD, which will have all the disadvantages of analog and digital, and advantages of neither. Your home made CD may well sound better than the factory CD, since the studio's air conditionar won't be filtered out.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Who said Nirvana? (none / 0) (#112)
by CaptainSuperBoy on Thu Sep 16, 2004 at 05:06:47 PM EST

I'm talking about any music that was made in the last 40 years, a time when studio recording equipment was better than consumer equipment. This means that consumer analog technologies such as the LP and the cassette destroy quality that is present in studio masters. Therefore CD is better.

--
jimmysquid.com - I take pictures.
[ Parent ]
Technical nit. (3.00 / 4) (#3)
by gordonjcp on Mon Sep 13, 2004 at 12:45:00 PM EST

RIAA compensation is not a kind of noise reduction. It's because if you move a magnet past a coil of wire, the faster you move the magnet, the higher the voltage you get. Therefore, a high-frequency signal will give a higher output voltage for a given deviation than a low frequency signal. Moving magnet pickups can produce some ridiculously high voltages at high frequency.

Ceramic pickups used to be common, and didn't exhibit this effect. They sounded crap, though.

Give a man a fish, and he'll eat for a day. Teach a man to fish, and he'll bore you rigid with fishing stories for the rest of your life.


Ceramics were awful (none / 0) (#30)
by mcgrew on Mon Sep 13, 2004 at 07:48:25 PM EST

an heavy. And true, it was not noise reduction, although it worked much the same way. However, it's the low freqs that bothered vinyl. Plug your standard stand-alone turntable into your sound card and you'll see what I'm talking about.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

not really (none / 0) (#91)
by tuj on Wed Sep 15, 2004 at 07:29:25 PM EST

The RIAA equalization curve is applied to vinyl recordings to keep the stylus from jumping out of the groove. On vinyl, the lower the frequency, the wider the groove in the record. Therefore, without attenuation, low frequency grooves would toss your stylus out of the groove. But your analogy is basically correct.

This also keeps the grooves narrower, which allows for more playback time on a side. Playback time on a side is directly tied to the width of the grooves. This is the reason your old LP's might have 15-20 min a side, and a modern dance 12" has 7-9 min per side. The dance records are cut "hotter" resulting in wider grooves, and better signal to noise, but less time on a side.

Believe it or not, this process does increase the signal to noise ratio of playback by reducing hiss and clicks. Since the high frequencies on the vinyl are boosted, the attenuation takes them back down while reducing the hiss sounds in the same frequency range. See here:
http://www.euronet.nl/~mgw/background/riaa/uk_riaa_background_1.html



[ Parent ]

why rip vinyl? (1.57 / 7) (#11)
by dimaq on Mon Sep 13, 2004 at 04:03:23 PM EST

you should rip the form/press/stamper they used to create the vynil. or the tape they used in mastering. or something.

ripping something you have bought in a shop just like millions other fools is just so uncool.

Fair Dealing (2.66 / 3) (#18)
by freestylefiend on Mon Sep 13, 2004 at 05:46:07 PM EST

In the UK, format shifting is not covered by fair dealing. It requires the permission of the copyright holder, so we break the law when we do this. This is not just the case for corrupt discs and was the case before the EUCD.

Why rip CDs, rather than just encoding what you capture from the cassette or LP?

It's less work (none / 1) (#27)
by mcgrew on Mon Sep 13, 2004 at 07:33:18 PM EST

You can record the whole shebang, and it only takes a couple of minutes to mark track starts. Then you can rip the CD to MP#, and you have the CD as a backup of both your original media and your MP3s.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Almost good (none / 1) (#22)
by 123456789 on Mon Sep 13, 2004 at 06:41:43 PM EST

Remove off-topic comments (such as your views of American and European nations, your views of best software to use, etc.) and you have a winner. I don't know who actually needs these instructions, but I guess k5 is a "general" audience, and I guess "general" includes liberal arts majors who need this kind of assistance.

---
People demand freedom of speech to make up for the freedom of thought which they avoid.
- Soren Kierkegaard
Views of best software to use? (none / 0) (#25)
by mcgrew on Mon Sep 13, 2004 at 07:30:04 PM EST

If I said that I'll delete it. Just because I personally like something doesn't mean it's the best. How should I word it to make it more clear?

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

I would remove this (none / 1) (#39)
by D Jade on Tue Sep 14, 2004 at 12:14:15 AM EST

This is because of a technology known as the "RIAA rolloff," which was a kind of "dolby" for turntables used long before Dolby patented the process.

This is untrue. RIAA Rolloff does exist but it is a specification to improve the quality of vinyl playback by using an equalisation filter to boost bass frequency response.

This kind of rolloff is still commonly employed in most low-end stereo systems and is designed to make the sound warmer and boosts the bass tones as most home systems do not have the capacity to produce bass signals below 20hz which affects the harmonics.

The real reason amplification is required is because you are recording from a phonograph and the crystal or magnetic styli which are most commonly employed produce very low level voltages.

Checkk out the Wikipedia Entry on this

In terms of Software, I would steer clear of recommending a package if you are not sure. A free sound program might do the trick. But if you really want power and control, you're going to have to spend some money. A cost effective choice would be a program like soundforge. It's a low-cost program which has the functionality to produce good quality results for simple recording projects like this.

I'd also compile a list of common applications for each OS which you can do by searching for Audio applications.

You're a shitty troll, so stop pretending you have more of a life than a cool dude -- HollyHopDrive
[ Parent ]

It's not egregious or anything (none / 0) (#48)
by 123456789 on Tue Sep 14, 2004 at 01:59:59 PM EST

... you handled it well. I guess what I should have said was that I'd like the article more if it was software-tool independant.

This is really just a nit I guess, becuase you have to have an example so that the article makes sense.

+1 anyway - good job!

---
People demand freedom of speech to make up for the freedom of thought which they avoid.
- Soren Kierkegaard
[ Parent ]
Next HOW-TO: (2.50 / 4) (#23)
by Esspets on Mon Sep 13, 2004 at 07:10:57 PM EST

HOW-TO convert from diapers to toilets.


Desperation.
Good idea, I'll get to work on it (none / 0) (#24)
by mcgrew on Mon Sep 13, 2004 at 07:28:09 PM EST

My kids were easy to train, though

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Hush up, you. (none / 0) (#59)
by Esspets on Tue Sep 14, 2004 at 11:45:18 PM EST




Desperation.
[ Parent ]
What is a good program for doing this under Linux? (none / 0) (#32)
by jope on Mon Sep 13, 2004 at 08:15:43 PM EST

Would have been interesting to hear about good free programs that do this and run under Linux or maybe MacOS.

Simple (sort of) (none / 0) (#129)
by CheezyDee on Mon Sep 20, 2004 at 11:07:10 AM EST

I'd use record (the other program that somes with sox and play) to record to a raw file, then use cat or dd to edit out any blank spots, and sox to convert to wav. I'm sure there's a GUI utility to edit audio tracks (perhaps cinelerra) but I've never used it.

I was thinking about ripping my LP collection a while ago, and I already have a decent turntable, but I get a LOT of clicks and pops due to dust and static. I need to find one of those Ronco record cleaners.

[ Parent ]

dd or cat (none / 0) (#137)
by nicolas e on Sun Oct 03, 2004 at 02:18:31 PM EST

How do you remove blank spots using dd or cat ?

[ Parent ]
I have a Nakamichi (2.00 / 3) (#34)
by CaptainSuperBoy on Mon Sep 13, 2004 at 09:40:49 PM EST

And I still don't rip tapes. It's just a whole lot of hassle. I have a manly 20 year old Nakamichi but I don't trust analog, and I don't trust my crappy onboard sound chipset. Even with a very good tape deck you can't hope to approach the studio quality of a CD. And with a poor tape deck you'll get a lot of hiss, of course.

Dead shows and other taper-friendly live shows aren't worth ripping because you can get shows via BitTorrent that were converted straight from DAT. There is no reason to introduce analog if you don't have to. As for albums, I guess it's up to you. I can't tell a CD from a 128kbps MP3, but I can hear the difference between a CD and a 30 year old LP of Blonde on Blonde that's been run through a $20 Crystal Sound chip with a mini plug.

--
jimmysquid.com - I take pictures.

pointless? (2.80 / 5) (#36)
by scatbubba on Mon Sep 13, 2004 at 10:00:18 PM EST

If the end result is to have music on your computer in mp3 form that matches your vinly collection, just go download the mp3s that you feel you have a right to possess.

Right to own a recording (3.00 / 4) (#41)
by D Jade on Tue Sep 14, 2004 at 12:20:07 AM EST

If you download a copy of a song from the internet that you have on vinyl, you don't automatically have the right to possess it.

If the copy you have downloaded has been digitally remastered, you do not have the right to own the recording. You may own the original recording on vinyl but you have not paid the producer for their work and are not licensed to use the re-recording.

Producers are artists too and have the same right to protect their work.

You're a shitty troll, so stop pretending you have more of a life than a cool dude -- HollyHopDrive
[ Parent ]

Conveniently ignoring the fact (none / 1) (#73)
by runderwo on Wed Sep 15, 2004 at 04:29:31 PM EST

that it's illegal for someone to distribute that file TO you that you are downloading without permission. Yeah, you're not infringing yourself, but if you knew it was illegal and the distributor on the other end was sued, perhaps it would be grounds for a conspiracy/contributory infringement charge. If buying goods, it is similar to whether or not you had the knowledge that the goods you are buying were stolen. In any case, you'll probably have to return them to their rightful owner, but if you knew they were stolen, you're in a heap more trouble.

[ Parent ]
Not always an option (2.00 / 2) (#88)
by mcgrew on Wed Sep 15, 2004 at 06:47:00 PM EST

Got a copy of Pink Floyd's Atom Heart Mother I can download? And I'm still searching Kazaa for most of Steppenwolf's Second (I've lost the vinyl).

And these were #1 selling bands in their day. Imagine trying to find somebody you like but who never sold many albums?

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

50% success (none / 0) (#94)
by gidds on Wed Sep 15, 2004 at 10:37:50 PM EST

Got a copy of Pink Floyd's Atom Heart Mother I can download?

AllOfMP3 has it.

And I'm still searching Kazaa for most of Steppenwolf's Second

I don't think they have that one, but they do have 11 other Steppenwolf albums...

Andy/
[ Parent ]

Atom Heart Mother (none / 0) (#127)
by Ni on Sun Sep 19, 2004 at 02:40:20 AM EST

ed2k://|file|Pink%20Floyd%20Ultimate%20Collection-22cds-Courtesy.of.EFnet's.%23B itTorrents.rar|1460073119|068761bedf756e47c0c53673e1e9ec49|/

This edonkey2000 file contains it. If this doesn't work out (it's a huge file), I could probably provide help in acquiring just that that album's MP3s were I emailed.


<mrgoat> I can't believe I just got a cyber-handjob from ni.
[ Parent ]

Software for Linux (3.00 / 2) (#43)
by Mtrix on Tue Sep 14, 2004 at 03:15:43 AM EST

I've been ripping vinyl for some time now under Linux, and here's what I use:
  • Kernel 2.4.something with low latency patches
  • arecord for the actual recording (I believe it's in the alsa-tools or alsa-utils package on Gentoo.) It should already be installed if you use ALSA.
  • Gnome Wave Cleaner to remove crackles and pops. You can find it here.
  • wavbreaker to split the big wav's into small wav's. Find it here.
  • normalize to get a similar volume as my other CD's when I finally burn what I ripped to CD (If I just burn them without normalizing they will have a much lower volume than my other CD's.). Get it here, or, if you run Gentoo, just "emerge normalize".
  • For burning I use K3b.
My soundcard is a Midiman/M-Audio Audiophile 2496. I've not yet tried with kernel 2.6.x, but I suspect I will get xruns. Maybe someone could enlighten me on how 2.6 will perform when recording.

Question: (none / 0) (#47)
by regeya on Tue Sep 14, 2004 at 01:23:44 PM EST

How do Gnome Wave Cleaner/wavbreaker and Gramofile compare?

[ yokelpunk | kuro5hin diary ]
[ Parent ]

i don't know about gwc (none / 0) (#49)
by cuz on Tue Sep 14, 2004 at 02:46:14 PM EST

but gramofile is just a low pass filter

[ Parent ]
I wasn't asking you, and you're dead wrong, (none / 0) (#72)
by regeya on Wed Sep 15, 2004 at 03:50:44 PM EST

but thanks anyway

[ yokelpunk | kuro5hin diary ]
[ Parent ]

GWC vs. Gramofile (none / 0) (#99)
by Mtrix on Thu Sep 16, 2004 at 04:48:49 AM EST

I haven't used Gramofile myself, but this is what "Converting 78rpm Records to Modern Media on Linux" (a good tutorial for ripping vinyl under Linux) says about them:

gwc -- The Gnome Wave Cleaner

This is a fairly new arrival in the field of Linux audio apps, but is the current "best-of-breed" for cleaning up sound files.
This is (as far as I know) the only available package that has all of the tools needed for processing the digitized sound files.:

Automatic click removal (with adjustable sensitivity).
Crackle removal.
Noise removal (several algorithms).
Manual click removal.


Gramofile

A curses-based program which includes both a recorder and a set of tick removal filters. As of this writing the current version is 1.6. In my processing, gramofile has new been largely superseded by gwc.


[ Parent ]
ecasound > arecord (none / 0) (#50)
by cuz on Tue Sep 14, 2004 at 02:52:17 PM EST

the final touches are being put on the 2.6 low latency patches and a massive migration of linux audio users to 2.6 is imminent.

[ Parent ]
Thanks! (none / 0) (#87)
by mcgrew on Wed Sep 15, 2004 at 06:43:49 PM EST

That was one reason for my writing this, was to get tips form you folks. I was especially looking for Linux solution.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

And on my Mac... (none / 0) (#93)
by gidds on Wed Sep 15, 2004 at 09:59:16 PM EST

I use a shareware audio editor called Amadeus II. (Thoroughly recommended; I also use it for everything from creating sequenced compilation CDs to mastering concert recordings. Some of its built-in effects are better than Cubase's!)

I record directly into Amadeus from my Audiophile 2496 sound card, and then use it for mastering: noise reduction, fixing clicks &c, trimming, normalising, and anything else that needs fixing. Your most important tools here are your ears!

If it's going to MP3, then I save as AIFF and use lame for encoding, as that seems to give the best quality. These days, though, most of my music is in AAC, which iTunes encodes for me.

Andy/
[ Parent ]

24 Bit Recording (none / 0) (#130)
by freestylefiend on Mon Sep 20, 2004 at 12:33:57 PM EST

Is your soundcard capable of 24 bit recording? If so, you should record in 24 bits and normalise it before converting to 16 bits. I used to normalise sound that I had recorded, but I decided that it wasn't worth the effort with my 16 bit soundcards.

[ Parent ]
I call bullshit (3.00 / 11) (#44)
by xL on Tue Sep 14, 2004 at 03:48:24 AM EST

You were almost making the right point when you mentioned that LPs can sound better because the recordings were mastered for LP. Then you screwed it up by mentioning the "better frequency response" of tape and vinyl. Get a grip, they generally don't. Both tape and vinyl suffer the same problems:
  • Analog media add dynamic compression to the audio. This is why the same master recording sounds like it has more oomph when pressed on vinyl when compared to the version on CD. It took a while for audio engineers to realize this when the audio CD became a new medium. Modern records are mastered for CD, so extra compression is added at the end of the mix to compensate.
  • Higher frequencies get masked by noise. Audio engineers compensated for this by attenuating the higher ends of the spectrum when mastering. This is why CD releases of albums mastered for Vinyl sound "too sharp" or "too sterile": There's too much high to compensate for a deficiency that is not there.
The fact that you can rip this music and, according to your own perception, it still sounds better than the CD version should give you an insight into your fallacy: You just demonstrated that is not a limitation of digital audio that keeps you from this better sound, but rather the fact that the CD version lacks an extra step of sound processing that was part of the 'engineered' sound of the record: The record player. Digitize the output of this analog 'playback' and you have a digital recording that sounds "just as good" as your analogue original, ruling out any matters of "better frequency response" or "better dynamic range".

Yes (none / 0) (#52)
by CaptainSuperBoy on Tue Sep 14, 2004 at 04:01:24 PM EST

You are correct about the fallacy here. And I guess the stock audiophile response would be that "conversion to MP3 degrades the quality so severely, that the original source doesn't matter."

--
jimmysquid.com - I take pictures.
[ Parent ]
Also (none / 1) (#54)
by Sacrifice on Tue Sep 14, 2004 at 07:46:40 PM EST

Assuming "half volume level" is on an intensity and not logarithmic scale, you'd only lose 1 bit, not half your bits (if peak was 65535, and you set it to 32767 instead, you're using 15 bits). To lose half the bits, you'd have to scale the peak down from 65535 to 255!

[ Parent ]
Minor nitpick (none / 1) (#63)
by xL on Wed Sep 15, 2004 at 04:00:09 AM EST

At 16 bits the maximum amplitude will be 32767, not 65536. You are right that the 'bit depth' is related to intensity and that going for 50% of the volume only costs you 1 bit. It's not impossible to screw up digital audio, but indeed the parts where the signal gets really bad are at insanely low levels. The only mention of the dynamic rane of vinyl records I could google right now mentions 60dB. If we forget about things possibly getting even worse for vinyl due to its SNR, the softest sound that can get out of a vinyl record can still be expressed as a 6 bit 44.1KHz signal. Personally I've never run into a digital recording where there was any noticable loss of definition on softer parts.

[ Parent ]
It is a logarithmic scale. (none / 0) (#86)
by mcgrew on Wed Sep 15, 2004 at 06:42:47 PM EST

To double your volume, you need ten times the power.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Points... (none / 0) (#85)
by mcgrew on Wed Sep 15, 2004 at 06:41:07 PM EST

* Analog media add dynamic compression to the audio

This is necessary because both cassette and vinyl lack the dynamic range of reel to reel (and CDs as well)

It's a completely different issue than frequency response.

# Higher frequencies get masked by noise.

You've neever actually heard a good turntable, have you? I suspect you've never heard a turntable at all. The noise is negligible. With a turntable, low frequency noise was usually more of a problem ("rumble") on cheap equipment.

Now, you misunderstood- I didn't say that your rip of a CD of a sample of a Pearl Jam cassette will sound superior. If the original master was digital, a digital copy will sound best.

However, there are a few "remastered form analog" CDs that are so inferior to the vinyl that sampling your own will sound better. Presence I mentioned, but the rest of Zepplin's discography, although better on vinyl, sounds better on factory CD than sampled from vinyl.

Many analog masters obviously were shit- Aerosmith's first album, for instance. You can hear tape hiss on the LP. Their second album was made with more care. Pink Floyd dropped one of their earlier labels over recording quality issues.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Compression (none / 0) (#121)
by xL on Fri Sep 17, 2004 at 02:40:18 PM EST

Tape itself has a compressing effect as well, compare what you send into the input of even a studio quality tape recorder with what comes out and there will be some compression. For vinyl, it may be a different story. I would at least expect there to be some dampening of a signal having to do with inertia, but I must admit that I haven't really listened to a high end record player to know what exactly happens to the sound.

At the very least, I seem to recall that the last step before pressing to vinyl included some postprocessing to make sure that the grooves wouldn't be cut too thin and the needle wouldn't be flung off the track, which implies dynamics processing that probably wasn't part of the mastering process for a CD release.

You do agree, though, that a proper sample taken from a proper vinyl cut played back on a proper deck will not sound 'worse' or 'more clinical' in a blind test, I assume? That's basically the only point I really wanted to make. If there never was a decent remaster made for CD and you prefer the way the vinyl album sounds, you have my blessings. Just don't go through life thinking "digital bad, analog good" because of this, that would be a misconstruction of the situation.

[ Parent ]

For those on OS X (none / 0) (#51)
by caek on Tue Sep 14, 2004 at 03:01:12 PM EST

Wanting to do the same, this excellent series of articles on ATPM will be of interest. (Links to parts 2 and 3 are at the bottom of the page, just above the comments).

Helpful (none / 0) (#84)
by mcgrew on Wed Sep 15, 2004 at 06:25:12 PM EST

Wish I could find good Linux tools, I'm sure they're out there.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Without any post-processing it'd sound terrible (3.00 / 3) (#53)
by nusuth on Tue Sep 14, 2004 at 06:36:39 PM EST

Record with highest sample rate possible on your sound card. If that is lower than 32 bit floating point samples, 48kHz sampling rate, upsample it. Slice to tracks. DON'T erase silent parts yet. Your tracks should start with the familiar hiss of old cassettes.

Get tryout version of adobe audition. If you have cooledit 2000 or cooledit pro, those will do too.

(It has been a while since I've done this so I might be misremembering names of options and menus.) Eliminate DC bias. Select the initial hiss part and get the noise profile, then reduce noise on whole recording using that profile (you might have to RTFM.) Now you can get rid of silent parts;  audition can do that for you. Eliminate clips if you failed to set volume just right and normalize volume to something sensible like %95. Frequency response is already fucked up, so equalize per taste (IIRC there was a preset for cassette digitization.) Downsample to 16 bits & 44kHz. Enjoy.

If you are too lazy, forget about working with each track one by one, upsampling, normalizing, equalizing and downsampling. Just eliminate bias, get noise profile from somewhere in the middle of the recording and reduce noise. Well worth the effort.

PS- (none / 0) (#83)
by mcgrew on Wed Sep 15, 2004 at 06:24:33 PM EST

Those functions are in EAC as well. With the ex's laptop I found DC bias correction very valuable, especially when recording my daughter's clarinet recitals.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

digress (none / 0) (#90)
by tuj on Wed Sep 15, 2004 at 07:12:13 PM EST

Ok, I'm going to slam this one. Yes, logically, it seems like it would make sense to record at the highest bit-rate and sample-rate possible. However, if the destination of your audio is CD, then this is not so wise.

Why? Because now you introduce dithering and down-sampling processes into the conversion to a CD-recordable wav file. These processes will nearly always introduce more artifacts and impurities into the sound than if you had just recorded at 16 bit, 44.1kHz to begin with.

Most dithering processes are noise-shaped, meaning that in trying to scale the 32bit data you've got into a 16bit range, there is an error that has to be accounted for. You could round, or truncate, but both of these solutions are worse than using noise to essentially spread the error out. This introduces background hiss.  (see: http://www.mtsu.edu/~dsmitche/rim420/reading/rim420_Dither.html)

I do agree with the noise reduction part (highlight a piece of noise 3 or 4 seconds long, use that as the profile.  Snapshots in profile = 900. Use a FFT size of 8192 or 16384, precision = 9).  I don't agree with the DC bias adjust (this usually causes more problems than its worth).  Its also important to remember that every time you normalize, you are causing inaccurracies in the data.  So only normalize once, and using a peak method to -0.1dB.



[ Parent ]

Dithering errors vs. roundoff errors (none / 0) (#101)
by nusuth on Thu Sep 16, 2004 at 08:06:45 AM EST

You might be right about sampling frequency. I've never given it much thought so I always upsampled both sampling frequency and bit depth.

I'm pretty sure that using higher number of bits while mastering sounds better than doing all noise reduction, equalization etc. at target depth though. The link you provide explains why dithering adds noise, not that addditional noise when downsampling is a bad thing (quite the opposite, infact.)

All time domain processing must be doable with target rate but higher bits should be noticably better with frequency domain effects, especially if you are applying more than one effect.

Anyway, this is all numerical analysis stuff I remember from university and some amateur recording experience. I'm not saying you are wrong but I'm not convinced.

[ Parent ]

my experience (none / 0) (#107)
by tuj on Thu Sep 16, 2004 at 01:09:29 PM EST

Well, my experience has been that dithering produces noticeable artifacts.  I used to record music at 32 bits, 48kHz, and after processing, dither and downsample down to 44.1kHz, 16 bit to burn on CD.

I started to notice that the improvements in the 32bit version I heard A/B'ing the same music recorded at both rates/depths not only dissappeared in the dithered version, but the dithered 16-bit sounded noticeably worse compared to the recorded 16-bit.  Note: I was using CoolEdit Pro to do my conversions. So my technique now is to try to get as hot as possible in 16 bits, and maximize the range, then process as few times as possible.

Don't get me wrong, the 32-bit versions always sounded better, and it was pretty easy to tell them apart from the 16-bit.  But it became useless for me to record at that depth when that added detail was not only lost on the conversion, but artifacts were also introduced.

Here's a good link on the dithering distortions caused by various commercial products:
http://audio.rightmark.org/lukin/dither/dither.pdf

Of course, if you are doing a lot of processing (say more than 3 operations) or processing that uses lots of multiplication (like convolution), I'm sure 32bit would be an advantage.  Or if your output is analog, higher is better.

I think what we really need is a popular format with better depth than the CD.  And while they're at it, make it the size of a mini-disc, enclosed in a protective case.  They should have done that with DVD's...

[ Parent ]

DC bias & noise reduction (none / 0) (#105)
by nusuth on Thu Sep 16, 2004 at 12:47:46 PM EST

I forgot to answer your point about DC: If you have a (real, not just statistical fluke) DC bias and your wave is not centered, noise reduction function will not work properly. It will create something like anti-noise profile noise in silent sections.

Also, I've never heard any difference between two signals differing only in DC (yes, I live in an apartment) what makes you say "...DC bias adjust (this usually causes more problems than its worth)"?

[ Parent ]

re: dc (none / 0) (#128)
by tuj on Sun Sep 19, 2004 at 03:14:54 AM EST

Well, I've rarely had a rip of a cd or record that had a DC bias problem, probably because of the medium itself takes care of that inherently.  And I can't say that I have had a recording of audio that has had its DC center above or below 0.  To me, DC bias adjust is only useful if your soundcard is so bad that it can't properly reference 0v.  

Now, the reason I say its more trouble than its worth to adjust is that unless the problem is inherent in your soundcard or setup, the DC bias problems will not persist thru the whole recording.  Meaning some points will have bias problems and others won't, and this fluxuates on a spectrum, so its rather hard to just adjust a portion without introducing distortion.

I'll give another example.  In the music I record, I use a significant amount of boost in the midrange on kick drums.  Combined with hard knee compression, this has the effect of creating a DC biased spike in the wave that exists only on the positive side.  The rest of the audio is fine, so trying a DC bias adjust just introduces distortion into everything else.

The answer, at least for me, is using a mastering program that uses various compressor and multi-band limiter models, which ends up fixing this problem.

But I do agree that if you record a wav file, and truncate the negative half, its surprising that it doesn't sound all that different (there is a difference in sound, but its not as dramatic as you'd expect, IMHO).


[ Parent ]

I disagree with most of this. (none / 0) (#114)
by awgsilyari on Thu Sep 16, 2004 at 05:44:14 PM EST

Most dithering processes are noise-shaped, meaning that in trying to scale the 32bit data you've got into a 16bit range, there is an error that has to be accounted for.

And there is NO error when the ADC samples directly from infinite-bit (analog domain) to 16-bit? Are you suggesting the ADC has some magical properties which it allow it to do this conversion without error?

In fact, the ADC also must choose which value to round the signal level to. It introduces error. By claiming that software cannot equivalently downsample 32-bit to 16-bit, you are essentially claiming that the ADC cannot be simulated by software. That's somewhat interesting, considering the ADC is a digital device.

Its also important to remember that every time you normalize, you are causing inaccurracies in the data.

Inaccuracies? That implies that the data was "accurate" to begin with. But we already know that the sound has been sampled from infinite-bit down to some finite number of bits, so there is no such thing as an "accurate" representation anymore. All that matters is the aesthetic quality of the sound. This tends to go hand-in-hand with mathematical purity, but not always.

--------
Please direct SPAM to john@neuralnw.com
[ Parent ]

ADC is a digital device? (none / 0) (#116)
by Meshigene Ferd on Thu Sep 16, 2004 at 06:03:40 PM EST

I think I don't like digital devices which introduce unpredictable errors. No, don't like them at all.
--
‮‫אַ גויישע קאָפּ!‮


[ Parent ]

Hello... (none / 0) (#118)
by awgsilyari on Thu Sep 16, 2004 at 07:45:50 PM EST

Welcome to physics, where there is no such thing as the "true value" of a signal.

The sample error is somewhat unpredictable because the universe is, well, somewhat unpredictable. Complain all you want. Won't help.

None of this has anything to do with the bit-ness of the ADC. The ADC uses a resistor ladder along with a small bit of digital logic to decide when to tip the scale toward one value or another. You suffer from the delusion that there is some way to do this which is "accurate." Sorry, the fact is, you are sampling analog data into the digital realm. By definition this introduces error.

--------
Please direct SPAM to john@neuralnw.com
[ Parent ]

Hi. (none / 0) (#119)
by Meshigene Ferd on Thu Sep 16, 2004 at 08:28:57 PM EST

You've just explained why ADCs are not digital devices. Which was my point all along.

Q: Wait a minute… so the grandparent post, in effect, says "ADCs are not digital"?
A: You are very bright.

Q: What D in ADC stands for, then?
A: I don't want to know. (Good) ADCs are made by analogue engineers, that's what counts.
--
‮‫אַ גויישע קאָפּ!‮


[ Parent ]

IAWTP (none / 0) (#134)
by garlic on Wed Sep 22, 2004 at 09:55:02 PM EST

As a digital engineer, I don't want to get any where near the design of an ADC or a DAC, especially with the amount of bits and dynamic range that's being required of the new parts.

HUSI challenge: post 4 troll diaries on husi without being outed as a Kuron, or having the diaries deleted or moved by admins.
[ Parent ]

accuracy (none / 0) (#120)
by nusuth on Fri Sep 17, 2004 at 12:55:53 PM EST

Inaccuracies? That implies that the data was "accurate" to begin with. But we already know that the sound has been sampled from infinite-bit down to some finite number of bits,

"Infinite bits..." So you imply universe is not quantified? Welcome to twentieth century, where energy is quantified, matter is energy in disguise, universe itself can't tell how fast you are if it knows exactly where you are and information that can be stored in a volume is proportional to surface area. You have been only 3 years late.

[ Parent ]

Re: accuracy (none / 0) (#122)
by glor on Sat Sep 18, 2004 at 01:53:46 AM EST

"Infinite bits..." So you imply universe is not quantified? Welcome to twentieth century, where energy is quantified...
I believe the point is that the precision of the signal is effectively infinite compared to the 16 or 32 or 64 bits available on the sound card, not that the universe is continuous. While there are circumstances where the quantum fluctuations in a signal are important (the technical term is shot noise), in analog audio these effects are masked by other noise sources like thermal noise, delay in feedback in preamps, noise from poor connections or cabling, and so on.

Also, most of us live in the twenty-first century.

While the rest of your post was correct, it was even less relevant.

IHBT

--
Disclaimer: I am not the most intelligent kuron.
[ Parent ]

YHnBT (none / 0) (#123)
by nusuth on Sat Sep 18, 2004 at 07:42:44 AM EST

I believe the point is that the precision of the signal is effectively infinite compared to the 16 or 32 or 64 bits available on the sound card, not that the universe is continuous.

You are using the wrong yardstick. It doesn't matter how much information an analogue signal carries and how it compares to 16 bits, 64 bits, whatever. We never notice finite precision of an analogue sound signal by our senses, for all practical purposes it is infinite precision while it realy isn't. So the correct way to look at it is whether we digitized music sounds any worse than analogue one. It is our perception that counts, not real information content.

[ Parent ]

-1, obvious (1.50 / 2) (#56)
by I Am Jacks Severed Testicles on Tue Sep 14, 2004 at 10:10:42 PM EST

Plus it does not cover the post-processing necessary to ensure that your rips do not sound like utter shite.

Support our troops - buy W Ketchup!
I should have said... (none / 0) (#82)
by mcgrew on Wed Sep 15, 2004 at 06:22:48 PM EST

It assumes good media to start, as well as good playback equipment to record it from. If you have a dirty, scratched up record or a cheap 39¢ tape you recorded from a ten dollar recorder and you're sampling with your two year old's Fisher Price record player it's going to sound like shit no matter what, although there are things you can do to make a shitty digital sample sound not quite so shitty.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

WHAT the FUCK? Hold on here a damn minute (1.33 / 3) (#57)
by RandomLiegh on Tue Sep 14, 2004 at 10:16:17 PM EST

..are you seriously going to tell me that 128kb mp3s sound worse or the same as casettes? because I'm listening to mp3s of shit I have on casette and my ear says you're full of shit.

---
Thought of the week: There is no thought this week.
---
In analog, equipment is everything (none / 1) (#80)
by mcgrew on Wed Sep 15, 2004 at 06:19:20 PM EST

Not so much so with digital. If you are listening to a cassette player made in the last ten or twenty years, it does indeed sound like shit.

I bought a walkman a couple of years ago to sample my tapes, and threw the damned thing away. I wouldn't bought it if I could have read the spec sheet before I bought it. Upper limit of 3khz, which is terrible, AM radio has a better response!

However, find a good quality deck from the 1970s (check pawn shops and used record stores) and it will sound much, much better than even a 192 kbps MP3.

I just burbed a Doors CD I'm listening to w/ a cheap headphone CD player (300hz-15khz stated freq response). The first half of the CD is their Greatest Hits album sampled from a factory cassette (Elektra/Asylum did an excellent job on it) with the last half LA Woman copied from CD (a bit for bit digital copy). Can you tell a difference? Yes- but only the tinyest of differences.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Except... (none / 0) (#95)
by Zerotime on Wed Sep 15, 2004 at 11:31:12 PM EST

You can't copy a CD bit-for-bit.

---
"You don't even have to drink it. You just rub it on your hips and it eats its way through to your liver."
[ Parent ]
192kbps MP3 (none / 0) (#113)
by CaptainSuperBoy on Thu Sep 16, 2004 at 05:21:11 PM EST

However, find a good quality deck from the 1970s (check pawn shops and used record stores) and it will sound much, much better than even a 192 kbps MP3.

You really don't know what you're talking about with MP3's. Almost nobody can tell a properly encoded 192kbps MP3 from the real thing. Try it - rip your favorite CD (completely digital, if you want) to WAV and convert it to MP3 using Lame 3.96. Use 192kbps average bitrate. Hell, use constant bitrate if you want, you still won't be able to tell the difference. Do some ABX testing and tell me what sounds worse.

And saying that old turntables sound better than MP3's.. you're comparing completely different qualities. With turntables you're arguing the dubious "analog" quality as if you're one of the few remaining holdouts who really believes he can hear frequencies above 20kHz, and sampling rates of 44.1kHz make the music sound "cold." It's audiophile bullshit. Discussing MP3 quality has nothing to do with the analog/digital debate.. it has to do with perceptible differences in the music. Of which there are very few at high bitrates such as 192kbps.

These are facts: CDs sound better than vinyl, period. And almost nobody can tell the difference between an uncompressed file and a 192kbps MP3.

--
jimmysquid.com - I take pictures.
[ Parent ]

Can you verify that? (none / 0) (#126)
by dougmc on Sat Sep 18, 2004 at 01:46:40 PM EST

I wouldn't bought it if I could have read the spec sheet before I bought it. Upper limit of 3khz, which is terrible, AM radio has a better response!
Sorry, but I don't buy the 3 kHz figure. I'd expect even a voice recorder to have slightly better frequency response.

Looking up the specs of cassette Walkmans (Walkmen?) online suggests that they claim a frequency response up to 15 kHz. (Like this one -- which even claims up to 8 kHz for recording.) Of course, there's been many different models, but even so, I don't believe that any Walkman designed for music listening only claims a 3 kHz frequency response.

[ Parent ]

EQ (3.00 / 2) (#58)
by the77x42 on Tue Sep 14, 2004 at 11:03:32 PM EST

The most important part in ripping vinyl is the proper setting of the equalizer. I run my vinyl through my mixer and set the EQ on there before recording directly from the REC-OUT into my laptop. Usually the low-end needs a lot of beefing up for older vinyl and adjusting the trim properly can give it an edgier feel.

Your article doesn't address this and merely tells newbies how to rip their shitty Nirvana tapes, which they can easily download off BT or Kazaa. Post something us nerds can use. -1


"We're not here to educate. We're here to point and laugh." - creature
"You have some pretty stupid ideas." - indubitable ‮

Yes, if your cassette isn't very good (none / 0) (#79)
by mcgrew on Wed Sep 15, 2004 at 06:12:07 PM EST

Or turntable. Some tape decks over-dolby, and some turntables (or at least their crossovers) lack bass. But I've found that generally, recording at a flat setting will result in the best sample. You can always eq the cd if your tastes vary.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

To rip vinyl... (none / 1) (#60)
by EraseMe on Wed Sep 15, 2004 at 02:12:02 AM EST

...just do this.

Any reports on the audio quality available? [nt] (none / 0) (#65)
by Empedocles on Wed Sep 15, 2004 at 05:38:45 AM EST



---
And I think it's gonna be a long long time
'Till touch down brings me 'round again to find
I'm not the man they think I am at home

[ Parent ]
With 11dB SNR, who cares.. (none / 0) (#71)
by geekmug on Wed Sep 15, 2004 at 12:44:20 PM EST

How about you exercise some 2nd grade reading skills and read the page yourself..

However interesting the idea is, they said they only got 11dB SNR on a 78RPM record.. nothing to write home to mom about.

-- Why reinvent the square wheel?
[ Parent ]
How about you exercise some (1.50 / 4) (#74)
by Empedocles on Wed Sep 15, 2004 at 04:46:38 PM EST

elementary oral skills upon my turgid member?

---
And I think it's gonna be a long long time
'Till touch down brings me 'round again to find
I'm not the man they think I am at home

[ Parent ]
Sorry for responding, ass.. (nt) (none / 0) (#124)
by geekmug on Sat Sep 18, 2004 at 12:01:08 PM EST


-- Why reinvent the square wheel?
[ Parent ]
Small technical error (3.00 / 5) (#69)
by wji on Wed Sep 15, 2004 at 12:00:46 PM EST

If you record your album at half volume, you have essentially recorded it in eight bits rather than sixteen. You want it as close as you can get to zero without going over.
Not quite. 8 bits encode 256 distinct volume levels, 16 bits encode 16,384. By recording at half volume you use 8,192 of those levels, or 15 bits worth. To reduce your dynamic range to 8 bit equivalence, you'd have to record like 18 dB below max. Mind you, you might be able to find ranges almost that great in classical albums, though the gods of pop recording have in their wisdom determined that all passages of all songs must sound equally loud.

I think you have to be some kind of olympic record holder to tell the difference between 15 and 16 bit audio -- I can only just tell the difference between well-noise-shaped 10 bit sound and 16 bit sound, but I have below average hearing. Heck, unless I go out of my way to listen for swishy stereo sounds in the treble band, most 128kbit MP3 sounds CD-quality to me.

Disclaimer: I am not an audio engineer, so there's a fair chance that some or all of this comment is wrong.

In conclusion, the Powerpuff Girls are a reactionary, pseudo-feminist enterprise.

Sound isn't linear, it's logarithmic (sp) (none / 0) (#78)
by mcgrew on Wed Sep 15, 2004 at 06:08:43 PM EST

To double the volume you have to increase the wattage by a power of ten. Your 100 watt stereo is half as loud as a 1000 watt Marshall (so long as they're feeding into the same speakers).

So a half volume signal wouldn't be much more than an 8 bit signal compared to a 16 bit signal.

This is where you'll hear the worst aliasing- in the softest passages.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

But we're not talking about sound . . . (none / 0) (#92)
by mysteray on Wed Sep 15, 2004 at 09:15:49 PM EST

It's a signal +- 1 or 2 volts. Very little _power_ is involved in transmitting that signal from the component to the computer. Binary encodings are generally log base 2, so indeed a reduction to half the voltage range is a drop in 1 bit. I'd gladly give up that extra bit to avoid the more serious distortion that clipping will introduce.

[ Parent ]
Resolution, not volume (none / 0) (#97)
by GhostfacedFiddlah on Thu Sep 16, 2004 at 01:04:59 AM EST

Actually, I'm not sure if "resolution" is the right term, but it's what I'm using.  You can easily counter the drop in volume of one-half by doubling the volume from whatever mp3 player you're using.

8/16 bit sound describes the resolution of the signal.  44 thousand times a second (44kHz), a sample is taken.  The amplitude of the sound at this point is recorded as a number.  In 8-bit sound, the number is between 0 and 255.  In 16-bit, the number is between 0 and 65535.  The reason 16-bit sound is better is because more frequencies can be stored with the higher resolution.

When reducing the volume by half, your 16-bit recorder will never go above 32767 (half of its maximum).  Thus you are actually recording at 15 bits, and leaving the highest bit at 0 for every sample.

To normalize later, you can simply double the volume of the entire wav file.  But you will still be missing that sound resolution.  This isn't related to the way the brain interprets volume.  This is simply a mathematical fact (and nowhere *near* the crappiness of 8-bit sound - try recording at 8 bits if you don't believe me).

[ Parent ]

Excuse me ... (none / 0) (#111)
by cdguru on Thu Sep 16, 2004 at 04:53:21 PM EST

but you have this utterly confused. Better than the previous posters, but still confused.

16-bit samples do indeed have 65525 different values, but they go from +32767 to -32768. 8-bit samples have 256 different values from +127 to -128.

16-bit samples are stored as:

+----------+----------+
| 54321098 | 76543210 |
| ---- 16 bit sample- |
+----------+----------+

8-bit samples are stored exactly the same, only the 8-bits represents the most-significant 8 bits of the 16-bit value. As:

+----------+
| 54321098 |
| -8 bits- |
+----------+

So, the 8-bit samples just lack the precision of the 16-bit samples, not the range. You convert an 8-bit sample to a 16-bit sample by simply shifting it up 8 bits and inserting 8 bits of zero. This is different than adding 8 bits of zero at the beginning as some posters seem to have been indicating.

8 bit samples carry just as much volume information, but lack the "fine detail" of the other 8 bits that are present in a 16-bit sample. Also, with a 16-bit sample it is possible to represent much smaller values than with an 8-bit sample.A 24 or 32 bit sample works the same way - the low-order information is extended, not the high-order. It doesn't get any louder, but there is more detail and it can represent lower volume levels.

[ Parent ]

information (none / 0) (#115)
by nusuth on Thu Sep 16, 2004 at 05:59:11 PM EST

8 bit samples carry just as much volume information

Your pet definition of "information" doesn't count. With 8 bits, maximum volume of sound you can represent is only 127 times the minimum volume you can represent, which is about 256 times lower than what is possible with 16 bits. But you can, of course, amplify both the lowest and highest volume. That you can get the same volume with both representations doesn't mean both carry the same amount of information.

Also, with a 16-bit sample it is possible to represent much smaller values than with an 8-bit sample.

You could have also matched the lowest sound value representable and say "with 16 bit samples you can represent much bigger values" which would be exactly as correct and exactly as misleading.

[ Parent ]

Pretty much (none / 0) (#117)
by GhostfacedFiddlah on Thu Sep 16, 2004 at 06:24:00 PM EST

I'd say "utterly confused" is a bit harsh though.  All I got wrong was asserting that samples are unsigned integers.

[ Parent ]
Resolution calculations (versus your theory) (none / 0) (#98)
by Gerhard on Thu Sep 16, 2004 at 02:21:23 AM EST

Actually half range for 16 bit is 15 bit.
16 Bits = 2^16 = 65536
Half of that = 2^15 = 32768
Nowhere near half = 2^8 = 256

Just record it at good medium volume and normalise, nobody will notice (including you).

[ Parent ]
True (none / 0) (#108)
by wji on Thu Sep 16, 2004 at 02:05:05 PM EST

Yeah, people perceive volume (and most things) logarithmically. It isn't exactly clear when somebody says "half volume" whether that means 1/2 amplitude or 1/2 perceived loudness. I was assuming the first.

1/10 amplitude would use about 10-11 bits, which is perceptible even in the loud parts. Quiet passages would be almost lost in the aliasing noise.

We are both right :)

In conclusion, the Powerpuff Girls are a reactionary, pseudo-feminist enterprise.
[ Parent ]

Snap, crackle, hiss... (none / 0) (#70)
by omiKron on Wed Sep 15, 2004 at 12:38:58 PM EST

Hmm, no mention at all of cleaning up unfortunate skips, pop, crackle, hiss, etc? I'm certainly no expert but I was getting mighty ticked when I began to record my old wax to the computer and kept coming up with intolerable problems such as what I mention. I've not been able to find any kind of tutorial on how best to reduce this kind of thing and all the experiments I did with "vinyl noise" and other such plug-ins were completely unsatisfactory.

In the end I just started to use the pencil tool in sound forge and touched up important songs manually. Someone's probably going to call me on this for doing it the wrong way, but like I said, I never found a method that worked - if you've got one, please share. Until then, I'll describe what I've been doing... it's fairly simple - you load your file into soundforge and go zooming in on each little artifact you want to clean up. Getting the big ones is easy - smaller ones can be tough. Zoom in real close until you can see the actual points of the wave, bust out your pencil tool, and try to get the waveform in the affected area to look like similar areas nearby. Usually you can find a trend and just try to emulate the look of it. It helps to hold down the shift key when a pop occurs in a mostly silent area. Also zoom in on the wave height to get more detail of what kinds of patterns may be hiding in what appears to be near-silence.

Maybe this is bad, I don't know. I just end up with music that is listenable after I'm done. Some normalizing, noise reduction throughout, and clipping the song down to proper size, and I'm done. It's good to have a full binder of vinyl rips on CD handy in case anything happens to my collection, I figure.

If there's a better way to do this all, by all means, let me know.
MUTATE & SURVIVE

I haven't found good solutions (none / 0) (#75)
by mcgrew on Wed Sep 15, 2004 at 06:01:51 PM EST

As to hiss, there really is no solution at all that doesn't involve filtering the high notes. A tape I just sampled a few days ago, Pink Floyd's Atom Heart Mother, was a cassette recorded from an 8-track that was recorded from an LP. Hisses terribly. If I find a copy of the CD I'll buy it.

You've found the only good solution to pops and crackles that I've found. A scratch on vinyl only leaves a mark on a hundred samples more or less, so you can delete them with impunity. If anybody's found a good automated tool I'd like to hear of it.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

PS- (none / 0) (#76)
by mcgrew on Wed Sep 15, 2004 at 06:03:29 PM EST

I should have added, it's assumed that your original media (LP or cassette) is in good shape to begin with, and I should have said so.

"The entire neocon movement is dedicated to revoking mcgrew's posting priviliges. This is why we went to war with Iraq." -LilDebbie
[ Parent ]

Adobe Audition (none / 0) (#100)
by nusuth on Thu Sep 16, 2004 at 08:00:11 AM EST

Audition has functions that are supposed to clean that stuff but I never recorded anything from vinyl so I don't know how well those work. For hiss, I found generic noise reduction works better than hiss reduction.

[ Parent ]
BTW (none / 0) (#102)
by nusuth on Thu Sep 16, 2004 at 08:11:13 AM EST

I'm not affiliated with Adobe, nor like Audition much. I used to love Cool Edit, but it was later sold to Adobe. Audition 1.0 was still the same program, only rebranded but it wasn't targetting the same market anymore. I can't say if current 1.5 version is any good. I'm pretty sure future versions will be crap.

[ Parent ]
What about WavePurity? (none / 0) (#138)
by Kanak on Sun Oct 24, 2004 at 01:29:40 AM EST

I have recorded some cassettes of quite poor quality. The recorded product had a very irritating hissing sound along with "Clicks & Pops". I tried to clean the sound manually using soundforge, but i think it simply isn't worth the effort. Better option for me, was to use WavePurity (www.Wavepurity.de). I received OK sound. Only downside is the time it takes to clean sounds (it took me about 6-8 minutes for each song of about 3 minutes).

[ Parent ]
LOL (3.00 / 3) (#77)
by chbm on Wed Sep 15, 2004 at 06:04:01 PM EST

"CDs simply do not have the undistorted frequency response of a quality cassette recorded on a good tape deck, which can record frequencies up to 18 khz."

Good one. However, the fact you consistently write kHz wrong throught the article tells me you're just pissing at the wind and I'll refrain from mocking you.

-- if you don't agree reply don't moderate --

wait, it gets better (3.00 / 2) (#81)
by chbm on Wed Sep 15, 2004 at 06:19:28 PM EST

"Newer cassette decks sound terrible, some with frequency responses that don't go up past 3khz or down below 500hz."

And then there's the bits about aliasing, those are good too.

-- if you don't agree reply don't moderate --
[ Parent ]

other favourites of mine (none / 1) (#106)
by mikpos on Thu Sep 16, 2004 at 12:50:00 PM EST

  1. For some reason LP -> WAV -> CD -> WAV -> MP3 is necessary. LP -> WAV -> MP3 just won't do!
  2. You need to download a third-party program and explain its operation in retarded amounts of detail just to record from your bloody sound card.
  3. Your results will be as good as (or better?!) those from a professional sound engineer working in a studio.


[ Parent ]
This is only really useful... (none / 1) (#96)
by failrate on Wed Sep 15, 2004 at 11:38:45 PM EST

...if it's an album or EP that was never released on CD, which is actually quite a considerable number...
Voodoo Girl is da bomb!
Hugely inaccurate (2.85 / 7) (#103)
by Contact on Thu Sep 16, 2004 at 10:58:41 AM EST

There are so many technical inaccuracies in this article I don't know where to start. I'm shocked that it hasn't already been demolished by someone, and I honestly don't have the time to go through it point by point, but although the actual ripping instructions may be accurate, the "technical background" is near delusional.

Examples:

This is because in a few cases, excellent audio engineers took great pains to make the LP sound as good as it possibly could. When remastered, these exceptional quality albums must have the treble tones attenuated to minimize the aliasing distortion. This causes the bass to be too loud, so that, too, must be attenuated.

Totally incorrect. Vinyl is actually limited in the frequencies it can handle (deep bass can actually cause the needle to "jump the groove", for example) while CD can provide a flat frequency response out to 20kHz+. CDs are often remastered when re-released to take advantage of the extra frequency range. That someone would take the vinyl master and roll off the treble and bass - I have no idea how you could have acquired such an idea.

CDs simply do not have the undistorted frequency response of a quality cassette recorded on a good tape deck, which can record frequencies up to 18 khz.

CD can go up to (theoretically) 22 kHz with a flat frequency response and minimal (> 90 dB) distortion, while almost no tape deck setups can match the frequency responses you're claiming here (certainly not any prerecorded tapes, which are normally run out of loop bins at ultra high speeds, and sound drastically worse than even the vinyl version).

Newer cassette decks sound terrible, some with frequency responses that don't go up past 3khz or down below 500hz. Even though its head will likely be a little worn, a used deck from the height of analog (1970-1980) will still sound better than the cheap junk they sell these days.

Is this a joke? 500 Hz - 3kHz is less than one octave! Such a tape deck couldn't even convey intelligible speech... this is a joke, right? I'm hardly a tape fan (listening purely to digital, these days, unless on a retro vinyl kick) but modern tape decks with technology like HX Pro will demolish the technology of twenty years ago. What next, you're going to recommend wax cylinders for the ultimate in high quality reproduction?

If you go over zero, you will introduce very, very ugly "clipping" distortion, just like recording to tape.

Actually, it'll be much worse than that. Tape has a certain amount of leeway when recording (effectively, it goes non linear), and small amounts of overloading can make the sound fatter (which is why some recording studios are still working with analogue tape for a warmer sound). Effectively it provides a simple form of compression (in the audio sense, rather than the data sense).

If you overload digital, on the other hand, the sound will be absolutely terrible, as digital has no leeway at all.

Some people put the volume well below zero. This is nearly as much of a mistake as letting it go over zero, as the lower the volume, the more aliasing.

Hmm. You don't actually know what "aliasing" means, do you? This is the second time you've used it incorrectly... aliasing occurs when you attempt to store a frequency which is more than half that of the sampling rate (ie - trying to record a 23 kHz tone on a sampler running at 44.1 kHz). It's effectively distortion in the frequency domain, and is indicated by atonal frequencies appearing in the spectrum below the midpoint. (If you're curious as to why, look up Nyquist's theorem).

You're correct that you want to get as loud a signal as possible to disk, but that's because the more bits that are used, the higher the potential signal to noise ratio. A 16 bit CD has a potential SNR of 96 dB - if you only record up to 25% of full volume, you're effectively wasting two bits, reducing your SNR to 84 dB. It's worth bearing in mind, though, that neither vinyl nor tape come anywhere close to 96 dB anyway (normally well under 70dB), so if you're using those as a source, it doesn't matter that much.

Your article was quite a good introduction to the techniques of recording, but I wish you'd left out the (almost all wrong) technical information.

Octaves (none / 0) (#109)
by phliar on Thu Sep 16, 2004 at 02:18:29 PM EST

Is this a joke? 500 Hz - 3kHz is less than one octave!
A thinko? 500Hz - 3kHz is more like 2.6 octaves, not less than one. [log2(3000/500)]

Although those frequencies are suspiciously high. The A above middle C (the standard note that orchestras tune to) is 440 Hz, and that's right in the middle of the treble clef. Middle C is 262 Hz (256Hz for quick calculations), and High C (double C) is around 1kHz — two ledger lines above the staff, that's one of those really high notes that break glasses in the cartoons. (Of course the harmonics will go way above that.)

Faster, faster, until the thrill of...
[ Parent ]

it's not all /that/ high (none / 0) (#110)
by Battle Troll on Thu Sep 16, 2004 at 04:29:24 PM EST

A piano's top C is two octaves higher.
--
Skarphedinn was carrying the axe with which he had killed Thrainn Sigfusson and which he called 'Battle Troll.'
Njal's Saga, ca 1280 AD
[ Parent ]
intelligible speech? (none / 0) (#125)
by dougmc on Sat Sep 18, 2004 at 01:30:23 PM EST

Is this a joke? 500 Hz - 3kHz is less than one octave! Such a tape deck couldn't even convey intelligible speech...
Somebody else has already covered the `size of an octave' thing, so I won't, but I do believe that 0.5 - 3 kHz could convey intelligible speech.

A narrow SSB amateur radio (ham) band radio filter generally lets in about 2 kHz of bandwidth. With SSB, 2 kHz of RF bandwidth corresponds to 2 kHz of audio bandwidth, and the voice is quite intelligible.

Granted, 2.5 kHz of frequency response is seriously crappy (I think telephones do around 8 kHz) but it should be at least capable of doing understandable speech (but probably not much more.)

[ Parent ]

THIS is incorrect. (none / 0) (#132)
by Shubin on Tue Sep 21, 2004 at 05:25:46 AM EST

>CD can go up to (theoretically) 22 kHz > with a flat frequency response This is incorrect. Theoretically it can not... unless you assume the existence of an ideal filter. But the ideal frequency filter is impossible EVEN theoretically. > and minimal (> 90 dB) distortion Measuring distortion in db is a strange idea itself, but again it is incorrect. Well, it IS correct for a uniform sine wave signal, but it is not a kind of music I like best of all.

[ Parent ]
Don't Use Your Soundcard (none / 1) (#104)
by birdsong on Thu Sep 16, 2004 at 11:37:02 AM EST

My goodness, computers are crazily noisy inside, why would you want to use your soundcard to do this? It's going to introduce all sorts of noise. Get an external analog to digital converter and send the signal into digital soundcard which does not remaster input. Using the procedure outlined here is about as far from professional as you can get.

Allocating Disk Space Before Recording (none / 0) (#131)
by freestylefiend on Mon Sep 20, 2004 at 01:09:18 PM EST

I use GoldWave on Windows to record sound. I have tried other programs, but I return to GoldWave because it allows me to create a specified length of silence to record over.

This is particularly useful to me because I sometimes decide to record long stretches of sound when I don't have much free hard disk space. Sometimes, when using FAT filesystems, I have found myself unable to use all of my free space to record a single piece of sound. (Is this due to fragmentation)? In such cases, if disk space was allocated on the fly, rather than beforehand, then recording would stop without prior warning. When using GoldWave, I can address problems of insufficient available disk space before I start recording. (I also insist on recording to hard disk because it often allows me to recover irreplacable audio from temporary files after a crash during or after capture).

Does anybody know of an open source audio capture tool for unix that has this feature?

Several corrections.. (none / 1) (#133)
by mindstrm on Tue Sep 21, 2004 at 08:15:18 PM EST

There are some good points in here.. however, there are some rather bad errors.

I won't get into the Vinly -vs- CD Debate.. as it's sure to start a war, but it's safe to say that the claims made about vinyl here as being so vastly and obviously superior to CD are simply not true.  You claim that vinyl has such superior frequency response, and then in the same article mention how you need a special preamp for vinyl that enhances bass and attenuates treble... something with good frequency response does not NEED processing like that.  

Vinyl may be superior in some aspects.. but not like you've made it here.

Casette tapes are absolutely NOT superior to CD in any way, period.  They do not have the same frequency response or dynamic range... and this stuff about a new tape deck having a response of 3khz?  you woudln't even be able to understand simple speech on that... that's bs.

CDs have a flat response all the way up to around 22Khz... and tapes have only 18Khz, as you said yourself... and CDs have far better dynamic range. how are tapes better?

Recording something at half volume in digital is not the same as recording at 8 bit.  Half volume would be 15 bit instead of 16 bit... remember, eacha bit doubles the amplitude.

Aliasing - I do not think this word means what you think it means. Aliasing happens when you try to record a frequency that is more than half the sample rate.

Further -  You should mention what audio settings to use for the recording.. what format, compressed or not, how many bits per channel, etc, and what the options are.

Further - How about some information on post-processing... there are good tools out there for digitizing vinyl that can remove pops and clicks, and you can remove tape hiss quite nicely.

An internal sound card is possibly the worst capture device possible.. if you want good recordings, get something external.

"This is because in a few cases, excellent audio engineers took great pains to make the LP sound as good as it possibly could. When remastered, these exceptional quality albums must have the treble tones attenuated to minimize the aliasing distortion."

Originals are on half inch tape, or similar... very high quality analog recordings, far higher than vinyl or casette tapes.  They do not have the rolloff and whatnot of vinyl.  Properly re-masterd onto CD this stuff will sound just fine, and arguably closer to the original than vinyl can get.
Why would they have to attenuate treble going from master tapes to 44Khz digital?  You use a 21Khz low-pass filter, of course, or you do get aliasing.. but you can't hear that stuff anyway...

Agreement in General (none / 0) (#135)
by virg on Fri Sep 24, 2004 at 10:07:59 AM EST

I agree with your points and sugestions for the most part, but I want to add a few comments.

> You claim that vinyl has such superior frequency response, and then in the same article mention how you need a special preamp for vinyl that enhances bass and attenuates treble... something with good frequency response does not NEED processing like that.

The preamping is to make up for the fact that the turntable won't put out the signal a sound card needs, not that the source signal is bad. Turntables that cost a pile of money (like modern DJ tables, for example) usually don't need this sort of preamping, but the goal of ripping off the platter is to save money, so what's the point of buying a quality turntable? His discussion addresses the gear that most people will be able to get cheaply.

> An internal sound card is possibly the worst capture device possible.. if you want good recordings, get something external.

See above. Most people own a plain-jane sound card already, so although it won't do as good a job as the higher-end stuff, you're saving money. At the cost of some of the external kits, you can get 50 CDs from the used CD store, so you'd need to rip a LOT of vinyl or tapes before you'd be saving a dime. Since cost is the issue for anyone interested in this, it's fair to suggest using the sound card you already own, although it'd be good to mention that there's better for the unusual case where it'd be worth it, like for out of print or really old recordings.

Virg
"Imagine (it won't be hard) that most people would prefer seeing Carrot Top beaten to death with a bag of walnuts." - Jmzero
[ Parent ]
Stale.. but here goes anyway. (none / 0) (#139)
by mindstrm on Tue Jan 11, 2005 at 07:34:08 PM EST

What comes off the needle on the record player needs to be adjusted for the RIAA compression that takes place on the low frequencies.  Low frequencies are compressed, and will not sound right unless the proper filters are used.  This is because high amplitude low frequencies will not fit in the width of the groove on the record unless compressed.

Your high end players may have this built in.. but a proper phono preamp does more than just boost the signal across the board, it also compensates for RIAA.


[ Parent ]

CD response 'flat' (none / 0) (#136)
by dcturner on Sun Sep 26, 2004 at 05:34:14 PM EST

Any digital recording, CD included, must at some point go through an antialiasing filter. In reality, this means the response of a CD is not 'flat' to 22kHz, but depends a lot on how good the antialiasing filter was. Recording at 48kHz gives you the opportunity to get away with a cheaper filter, but it's never going to be perfect.

Remove the opinion on spam to reply.


[ Parent ]
Antialiasing filter? (none / 0) (#140)
by mindstrm on Tue Jan 11, 2005 at 07:37:56 PM EST

It goes through a low-pass filter before being digitized, yes, and a crappy low-pass filter is going to affect things.

This is why sampling is often done at much higher freqneuces and lower bit widths, and then converted later: so that the low-pass filter can be very, very, very gentle.

Further, we are talking about what is recorded, not what the original signal was.  Analog is often filtered as well.


[ Parent ]

How to rip from vinyl or tape | 140 comments (112 topical, 28 editorial, 3 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!