Audio Cassette Conversion

[wyclef]

Hey, i'm trying to convert old audio cassettes to good quality audio to try to polish up. I have an old 4 track and was able to hook it up to my macpro and record the audio playing in adobe soundbooth cs5 and first try was ok but just want to get some advice on how to do this properly to get the best recording. It seemed as though i was having some trouble with the decibels maxing out in the recording. Tips? Advice? thx.

[filmboss80]

Make sure your audio cassette deck's outputs are plugged into the Line In of your computer's sound port -- NOT Mic In. This may be why your levels are overdriving. Another cause for overly high levels is simply that your input volume is set too high.

Be sure to save to a lossless file type, like LPCM format .wav. (CD quality containing two channels of 44,100 samples per second, 16 bits per sample.) You can then filter, edit, and alter any way you please. Finished .wav files can be used for making CDs or converting to .mp3 or other format.

[wyclef]

what's up with 16, 24, and 32 bit?

[filmboss80]

What is your targeted output? CD, DVD, or something else? Bitrates of 24 and 32 are not compatible for CDs. (See http://24bit.turtleside.com/ for more info.) Plus, if your source is audio cassette, there is no point for the higher bitrates.

[wyclef]

not sure. just trying to salvage some degrading audio cassettes and get a good recording so that i can decide what to do with them another time so i'd like something so i won't have to redo it again cuz i think these tapes are on their last leg.

[Cornucopia]

Not true, filmboss80,

Yes, for CD compatibility you want your FINAL, EDITED WAV to be 16bit, 2ch, 44100Hz, LPCM (and for DVD compatibility, to be the same, but at 48000Hz, or compressed to AC3), but assuming the hardware card and the software support it, it makes very much sense to capture at a higher bitdepth and higher sampling rate.

All NR, FX, and mixing do minor degradation to the signal. If you want to END UP with CD quality, your best bet is to start ABOVE CD quality.
Now, we all know that Audio Cassettes don't have greate S/N ratios, nor high bandwidth. But by being ANALOG, there is information there to be used, particularly if apply Normalizing and Noise Reduction.

Maximize your available bitdepth, using a >16bit mixer (I like ProTools' 48bit mixer, which has rediculous amounts of headroom) and using 16bit dither, so that the upper 16bits have the valuable data in them. Hopefully, then the rounding+truncating at export/render will have "all the good stuff" in it. Similarly with bandwidth and a final, phase-linear lowpass filter before export.

If you appreciate your audio, it makes a difference - EVEN for audiocassettes.

This kind of approach works similarly in the Photo or Video realms, as well.

Scott

[wyclef]

what would be the optimal dB output I would want to target? i between -18 and -6???? what do you do if you record 40 min of tape and then the dBs peak for a second? lower the levels and start over?

[Cornucopia]

If you set the RMS to be ~18db, then you basically have 18db of headroom for the peaks. Engineers use anywhere between 12-24db of headroom, usually 15, 18 or 20 being the most common. This is all program dependent anyway (classical live music would need more, a cassette tape of rock music should need less).
Then, if you've captured nominally at -18db, you're wasting ~3bits as the S/N ratio works out to ~6.02db per bit. Captured at 16bits, you're already at a disadvantage. Captured at 24bits, even if you're wasting 3 bits, you've still got 21 bits of SN (=126db). Plenty to spare.
Then, normalize all your clips identically (to keep the inter-song balance the same). Let's say that raises the level by 8db. Your nominal level is -10db, with some peaks @ ~1db and a few lower at say -5db.
When you're finished, your entire tape's contents is at optimal level without clipping. If you dithered the mix & nomalization, etc. You've raised the noise level by a smidge, and then removed in in the truncating to 16bit, leaving just the clean sound.

That's why I suggest you use higher samplerates/bitdepths. Finished sound is optimal.

Scott

[wyclef]

sorry, i'm kind of a noob here... i'm using adobe soundbooth cs5... is there any chance you could translate into a language i might be able to implement? i don't even know what RMS is? sorry

[Cornucopia]

RMS = Root-Mean-Square aka AVERAGE.

The idea is you set your capture bit depth to higher than usual (like 24). Then set your input level/gain/sensitivity to somewhat LOWER than usual (since you now got so much extra to spare). Then capture everything. Then, applying dither throughout, you - Edit, FX, and then Mix and in the final process, raise the level to JUST BELOW clipping for the HIGHEST PEAK in the ENTIRE tape/program. Then export to 16bit. Your programs would AVERAGE around -10db, with peaks around -1 to -5db (using the fictitious example cited above).
You could have used a compressor/limiter to raise/tighten the sound a little more, but I don't recommend it unless you know what you're doing and you have to do it (like prepping for radio stations).

Scott

[wyclef]

what program do you use?

[Cornucopia]

Whatever gets the job done. I probably prefer ProTools, if the supporting hardware is available. I've used (listing audio editing apps only here): PT, LogicAudioPro, Cubase, CoolEditPro/Audition, Audacity, Goldwave, Reaper, SoundForge, Sonar, Samplitude/Sequoia, Reason, SAW, SoundtrackPro, Soundbooth. (Undlerlined ones I use more often)

Your Soundbooth is a modified subset of CEP/Audition, with Task-based toolset. Meant for Video folks who need to occasionally dip into Audio.

Scott

BTW, filmboss80's comment about Line Level Ins and Lossless/Uncompressed formats is spot-on.

[wyclef]

wow, i just downloaded the beta of adobe audition. so much better than soundbooth!

[wyclef]

ok, i've uploaded a screen. need more pro eyes on this... i'm frustrated because i can't seem to get the bulk of the audio recorded at a higher level because there are little bits here and there that get clipped. do you think this looks sufficient for old basement band recordings on cassette i'm trying to digitize and improve?

[Cornucopia]

I would say if you only have a handful of spikes (pops, ticks, clicks, glitches) and the rest is fairly even level, don't sweat it (bring up the level in capture/recording). Then after-the-fact, manually remove those spikes (through editing/replacing waveforms, manual microscopic waveform redrawing, or more automated pop/tick removal dsp/filters).

If there are more than a handful, keep the level lower (like I said earlier) during capture/recording, THEN remove the spikes, THEN raise the level.

Once again, a good reason to use higher bitdepth. Of course, your hardware has to support it as well (I was assuming that, but thought I'd better mention it just in case).

Scott

P.S. Other than the spikes, the waveform looks fairly decent; no real clipped sections.

[davexnet]

To get the benefit of 24 bit capture, don't you need a sound card with true 24 bit ADC ?

Or, if you only have a 16 bit ADC, is there any benefit to converting to 24 bit in the audio editor before any further
manipulation is done ?

[Cornucopia]

No, you're correct, you DO have to have a 24bit capable card. That's why I said the "your hardware has to support it as well" part.
Although, usually the driver for the card WON'T let you choose higher bitdepths if the card doesn't support it.

Scott

[wyclef]

great, now i have to demagnetize and clean the cassette player

[Cornucopia]

??

Well, that's SOP for good xfers, anyway.

Scott

[wyclef]

yup, problem is i dont have the stuff handy... man, what was supposed to be an easy peezy project is turning into a huge time killer here.

[bendixG15]

Cornucopia - Thanks for the info