High Speed Audio Cassette to MP3 Conversion? 17
tottydoc asks: "I have
a few thousand cassette tape recorded lectures (monoaural) that I
wanted to convert to MP3. Are there any high speed dubbing decks to
use in combination with some software to do this quickly?"
Might there be some quality concerns when recording audio to disk
and then slowing it down to the intended speed?
Elbow-grease ware. (Score:3, Funny)
Suggestions (Score:4, Funny)
1. Hire a neighbor kid to sit at your computer while you're at work and do this, tell him you'll give him $40 and buy him a pizza if he'll change tapes and watch TV all day.
2. Use Musicmatch JukeBox.
There simplying isn't a good wat todo this without losing a lot of quality. Sorry =( but hey if you do come up with a way todo it, let us know!
Fiddle with the sample-rate... (Score:5, Informative)
Then just grab the signal from the playback-deck and feed it in your soundcard. Record it to a
Now the encoded mp3 should have the correct speed and frequency again.
Of course that would limit your Bandwidth to about 5000 Hz but I'd give that method a try to check the feasibilty of that method.
If it works out halfway useable then you could try and get one of those 96kHz/24Bit-soundcards, that should yield about 12kHz Bandwidth (96kHz recording, 24kHz encoding -> 12kHz BW) which should be more than what you could expect from a normal analog audio-tape.
Of course this totally ignores any frequency dependant effects in the path from the magnetic media to your computer... you may have to compensate these with some filters... maybe there are encoders that have an equalizer simmilar to the mp3-players?
To check your recording's quality I'd recommand baudline (http://www.baudline.com/)
Re:Fiddle with the sample-rate... (Score:1)
Say you have a good recording card with 4 stereo inputs, thus 8 channels. You could setup some software to record all 8 channels separately and send them off to the encoder in real-time on a 900mhz Athlon. Combine this with the speed-doubling or quadrupling technique, and you could be ripping at an effective 32x speed once the whole thing is decently setup. Of course you'd then need 8 tape decks, but that's a minor issue I think, given the size and importance of your audio archive. It would reduce your ripping time from several months to only a few days, assuming you're doing this 8 hours a day (or someone else is doing it for you).
distributed effort (Score:1)
find yourself 40 tape decks, and either use them at normal 1x speed, or use the above solution for cutting your sample rate in four to accomodate a 4x playback.
i recall either a built-in feature or some plugin that allows you to split a wave file into segments based on noise thresholds, just set this to maybe 30 seconds of silence to split the tracks up.
or, you could set up a macro program, and bind one key to save an existing wave file, and open a new one. bind this to 4 keys, and you'll easily be able to control 4 streams.
assuming that you have a thousand tapes, thats about 40 days worth of playtime, and so this method should take 1/4 to 1 day to complete the ripping. as for the encoding, use some mass encoder. you could even use a network share and write a script to encode when a file isnt written to anymore.
even better yet, find one big lecture class, and pass them out as extra credit, for anyone who'll rip them.
Re:distributed effort (Score:1)
Re:distributed effort (Score:2)
A few thousand? Ouch. (Score:4, Informative)
Since there are such huge quantities of material, I can only assume that quality is of little importance, as long as the end result is plain and intelligible. Alas, you failed to mention any budget or quality concerns...
The most unavoidable loss with high-speed playback of cassettes is that introduced by the tape deck's playback head. These things don't deal with high frequencies very well at normal playback speeds - let alone after frequencies double or quadruple with tape speed.
That said, Ma Bell doesn't think there's any useful information in the human voice outside of the band of 300-3,000Hz, so they limit telephone calls to these constraints. I don't have much trouble understanding people on the phone. I'll use this as a baseline for intelligibility, and thus assert that 4x playback would be fine on all but the rattiest tape machines.
Any good service tech will be able to modify a cheap tape deck to play at high-speed. A machine already designed for high-speed dubbing would be ideal, as it makes the mods easier (just trick it into thinking a tape is present in the record well) and they sometimes have heads designed for playing very high-frequency sounds.
Tascam and others make decks intended for mass-duplication of cassettes. These will have 1 playback compartment, and a number (from two to lots) of record wells. IIRC, these can operate at 4x normal playback speed.
Record at 44.1, 48, 96KHz, or whatever sampling rate floats your boat. Any of these should be sufficient, with 96KHz being overkill for this application.
Next, lie to your MP3 encoder about the sampling rate - divide it by the playback speed. For example, if you record double-speed audio at 48KHz, tell the encoder that the file is 24KHz.
Encode the previous tape while you're recording the next. This multitasking will allow the tape machine to stay busy. If the computer put up to the task isn't very fast, avoid VBR and other CPU-robbing features in the MP3 encoder to make sure that things don't get backlogged waiting to encode.
It'd be possible to save more time using multiple sound cards and tape decks, given some good organizational tactics.
Since the tapes are mono, it would be almost trivial (with sox and a lot of scratch space, named pipes in real-time, or what have you) to record two at once, using the left and right channels of just one sound card. With poorer-quality cards, this will might have a bit of crosstalk. It's up to you if it's acceptable or not.
Another issue which might be important:
I doubt that the speed of most "high-speed dubbing" tape machines is calibrated at all. They generally use a single motor to drive both the record and playback tapes to ensure that the copy comes out at the same pitch as the original. Better machines sometimes have completely seperate mechanisms, but the better ones also typically forego the high-speed function altogether...
So, if the pitch (or elapsed time) of the end result is important to this project, it would be a good idea to ensure that a given deck's 2x mode -really is- 2x, and not 2.3x or some other random number. Again, a good service person will be able to help you here.
Last, as a previous poster suggested, pay a neighborhood kid in small bits of cash and pizza to do all of this. And have him dump the resultant mp3s onto CD-R for you. Twice. Put one stack of CDs far, far away (the next town would be nice), as you'll never, ever want to do this again - even if there is a horrible fire, followed by a flood, just before a hurricane drops a 747 on your lecture archives.
-
Something I'd like to see... (Score:1)
Cassette decks for PCs? Good Lord.. (Score:3, Funny)
- A.P.
Re:Something I'd like to see... (Score:1)
Re:Something I'd like to see... (Score:2)
I went to work for a publisher who was launching a new computer-hobbyist magazine. (No names, but rhymes with "fright".) In his office was one of those brand-new MITS Altair 8800s. It was running a blinkinglight routine, its binary string of lights simply counting upwards. Its main I/O consisted of the front panel lights, the front panel sense switches (something like 8 bits, address, store, and load next, IIRC), and a cassette connection.
Now to boot this puppy, you used the sense switches to load a 30ish byte bootstrap program into memory. One byte at a time, each bit set individually. Like an old PDP-8 or other primitive mini, which it was designed to resemble.
On cassette, we had Altair 4K Basic. But we couldn't figure out how to get it working. We did manage, however, to get the system to read in at least some of the tape. Then we read out the ASCII from the binary patterns on the lights and decoded the message. Something like this:
MICROSOFT ALTAIR BASIC COPYRIGHT 1975 BILL GATES AND MONTE DAVIDOFF. Then some binary.
So we called up MITS in Albuquerque and asked to speak to Mr. Gates. He got on the phone and explained how to read in the tape and start it executing. This let us then use a terminal like a Teletype 33 as an I/O device.
Who would have imagined?
(Okay, it's a digression, but it wasn't until about five years later that floppies became affordable to the high-end home user, so various incompatible cassette-input systems remained popular for several years. And were never much good. There were even "standards" for bits on cassette, like Tarbell and Oklahoma City.)
A good way to spend your stock-option money... (Score:2, Informative)