Hi,

I've started digitizing some of my cassettes recently, and while reading up on how to do so, I came up with a few questions about the process.

One of them is what the best software for the capturing the audio is. I know that Audacity is a good free program, but are there any commercial applications that are significantly better? I'm using Adobe Soundbooth right now - is that a good program for this purpose?

The other thing is that from what I understand, to make sure you get as full of a sound as possible from the recording without flattening the volume levels, you need the waveform to get as close to 0 dB as possible on the loud parts without going over it. Is that correct? Here are the waveforms from two sides of a cassette that I digitized. Which one would you say had better volume selection on the cassette player?

Side A:
http://img527.imageshack.us/img527/4685/so...veform02pi2.png
Side B:
http://img47.imageshack.us/img47/1887/soun...waveformkn5.png

Thanks a lot!

For recording, unless you select a defective program, every application will produce identical results. The only real variation is between 16 and 24 bit input; some programs will not handle 24 bit so you will get 16 bit, regardless of what the soundcard is producing. If you intend post recording process, there are many variations; some programs will be much better because they have better facilities and do more accurate processing.

There isn’t much point to 24 bit for cassettes, you will not get any more information than with 16 bit, nor will you be able to tell the apart. 24 bit (or better, floating point) is useful if yo intend to do a lot of post recording processing, but you will have to do many steps to have a chance to tell any difference. If you need it, you can convert a 16 bit recording before you start the post recording processing.

If your peaks are with 6 to 12 dB of 0dBfs, they are fine, no need to worry about getting a higher input level. The first image looks like clipping is occurring, which is worse than the potentially lower signal to noise ratio from a lower input level. If you have a decent soundcard, its noise floor will be so far below the cassette deck that there is no noise to be overcome.

How are you amplifying the input signal? How much noise does that add to the recording? If the equipment does not add significant noise, there is no downside to getting close to 0dB, as long as you avoid much clipping, but it is generally more pain and trouble than it is worth to have to fight inputs adjustments to get that “perfect” level.

First one is a _little_ hotter than I would do - maybe a dB but certainly OK. The second is a bit low but not serious. Keep in mind the noise floor of the cassette even with Dolby is much higher than the digital you're converting to. I would be more concerned with the head and EQ alignments of the cassette deck but that is beyond what most folks are equipped to handle.

What you're using now appears to be pretty good but I haven't used Soundbooth.

Personally I use Adobe Audition 1.5 for audio work including LP conversions to CD, MP3 files, noise reduction, level changes, channel mixing and some looping of sound samples for GigaStudio.

G²

For me, side B looks better. Side A looks like some clipping occured (for example, just before 20:00, at ~12:40, etc).

Unless I've missed something:
First one looks clippy to me and almost certainly not OK. The second one looks right on the limit (see 21.5 mins in) rather than "a bit low".

C.

I'm not sure what you mean by "amplifying the input signal. I'm using a boombox with a cassette player to play back the cassettes. I connected one end of a 3.5mm audio cable to the headphone output of the boombox, and the other end to the microphone input on my laptop. That should provide a good recording, right?


You're right - I read some posts about the head and EQ alignments contributing to the quality, but I have a pretty simplistic hardware setup, so I think that is beyond what I can do with the hardware I have.



The boombox that I'm using has a very limited volume control system, so I don't think I can adjust the input volume very finely. I think I'll redo the first one and see if I can improve it a bit, I suppose. On most cassettes, would listening to the first few minutes of the recording and determining if the volume is correct for that section be enough to get an accurate estimation of the optimal volume for the entire side of the cassette? That is, different songs on the same side won't have widely differing optimal volumes.



I have an Intel ICH8 Integrated sound card. That's not exactly a quality sound card, but it's fairly new. It should be OK, right?

Really depends what's on the cassettes; what kind of music (big dynamic range or not etc ...)

C.

This could be a major problem. The microphone input is probably much too sensitive for this application. If your laptop does not have a line input then you will probably have to seriously attenuate the signal from your boombox to not overload it, and even then you may be adding excess noise.

I'm just splitting the recordings into individual songs, and then encoding to FLAC. I'm splitting the recordings in Soundbooth, but I don't see an option to encode to FLAC in it, so I'm just using foobar2000 for that right now. If Audacity can do the FLAC conversion, that may be good enough of a reason to switch.


I'm checking the waveform of the recording to make sure there's little to no clipping. Shouldn't I be OK in that case?


Here's my second attempt at recording side A: http://img153.imageshack.us/img153/281/sou...veform03fs0.png

That's the kind of waveform I'm looking for, right?

If your laptop has no line input then I would look into a USB sound card with line input. The microphone input has higher distortion and less dynamic range than a line input.

How much would a decent card cost? Could you make some recommendations?

As others said, the mic input expects levels in the -40 range and has more noise issues than the line input (and even that is lame on most sound cards) but more importantly, they are often Mono. It's hard to tell for certain from the waveforms but they certainly look mono. Going mono would be a deal breaker for me if the original material was stereo.

G²

What sort of material is on the cassettes? Is it irreplaceable original music or voice, or is it commercially available music? If it is the latter, surely you would be better off finding digital versions of the same material.

Of course, if it is the former - carry on!

Check out these.

Audacity does indeed have inbuilt flac support: features

It's classical Indian music that probably isn't available here in the US, and I'm not sure when I'll be able to visit India next, or even if they sell these recordings in a digital format over there. I'll be sure to check online to see if I can find it though. Almost all the sites that offer downloads have lossy MP3s, and some of them even have formats like RealMedia or Windows Media.


I took a look at Newegg already, but are there any particular brands that I should look at, or can I just go by the ratings for each individual product?



I'll have to give it a shot then!

In terms of their line in, these sound cards may vary a little in noise level, but they should all be well below the noise level of your cassettes. If you plan on doing a lot of processing then higher sample rate or bit depth might be worth considering, but otherwise just go with 16-bit 44.1 kHz to be compatible with CD.

I would keep peak level a couple of dB down from 0 dB, to avoid intersample overs. I doubt your meters are good enough to provide 'true' metering of digital levels.