Folks, I wonder whether any of you can offer advice about digitizing a collection of cassette tape recordings. I have about 300 cassettes that I recorded live (and through the atmosphere, not straight from the sound board) at various dance events--some called by me, some where I was just another attendee--mostly during the 1980s and 1990s. I'd like to get them digitized and do some processing and indexing to make them easier for me to listen to and study. (For various reasons, I'm not currently looking to make them suitable for distribution.) I'm interested in learning about software for a variety of purposes: capturing audio signals as WAVE files; applying various kinds of digital signal processing (DSP), such as cutting some of the tape hiss and transport rumble; and labeling, indexing, and annotating, complete sound files and portions thereof. With 300 or so cassettes to deal with, any tools that make it more efficient to process them, to annotate and index them, and to keep all the files and notes organized, could significantly reduce the amount of time it will take, or significantly increase the amount I get done before I run our of steam. I'm looking for recommendations of programs that will run on a Mac (with OS X), and that are not insanely expensive. I'd also like to end up with files that are in common standard formats or that can easily be converted to standard formats without unnecessary loss of information. For those who want more detail, below are some rambling musings about things I might want to do and for which software assists might be helpful. * * * * * * * * * * * * For one thing, I'd like to do some noise reduction. I recorded all these cassettes using a variety of little Walkman-sized portable cassette recorders, mostly using the built-in mics, which pick up transport rumble from the rollers and spools. And of course they all have some amount of tape hiss, which is more or less evident depending on the volume of the actual program content on each particular tape. The transport rumble and tape hiss can make it fatiguing to listen for long periods, and I'd like to eliminate as much of them as I conveniently can without giving the sound an obvious "processed" feeling that would be similarly fatiguing to listen to. Another thing I'd like to do is to annotate the files with index marks for sections of the content. A typical live recording of a contra (and/or traditional square) dance will naturally divide mostly into segments of set formation, walk-through, and dancing to music, plus other miscellaneous stuff like couple dances and announcements. It would be nice to know of a program that would let me create index points, label them ("Walk-through for dance 3"), adjust them (retaining the labels), re-label them ("Walk-through for Double Boomerang"), skip directly to a chosen index mark during playback, etc. In some cases, I might want to apply long annotations, such as a detailed description of the figures and choruses of a square dance, which I might want sometimes displayed and sometimes hidden. I also might want to annotate sound files, or portions of them, with things like dates; names of callers and bands/musicans; venues; session titles; etc. (Of course I realize that I could also write such detailed notes in separate text files and use some sort of file naming discipline to keep them associated with the correct sound files.) In any case, it would also be good if I could conveniently get (accurate) times and labels of all index marks and other annotations dumped to a plain text representation. This would insure that (1) the data wouldn't be hostage to the proprietary format of some company that could stop supporting it or simply go out of business, and (2) I could search through the labels and annotations using ordinary UNIX tools like grep, which I trust to be reliable. I presume that it would be best to start by capturing the content from the tapes in WAVE format and to do any noise reduction or other processing on the WAVE files. Then if I wanted to create MP3 files to save space, I could do that afterwards, also saving the original WAVE files on some off-line medium. [Any suggestions about which off-line media are likely to have long shelf life?] Besides reducing tape hiss and transport noise, there are some other kinds of digital signal processing that might be useful. For one thing, I don't trust that the little battery-powered recorders I used did a great job of keeping the tape speed accurate, especially as the batteries began to run low. They might even have done a bad enough job to significantly affect any timings I might do of the musical tempos on the recordings. I can think of various ways in which digital signal processing might help determine accurate playback speeds. For example, a frequency spectrogram might reveal a small but detectable spike near 120 Hz that came from fluorescent lights originally humming at (almost) exactly 120 Hz. Spectral analysis of the music might give a clue about whether it's being played back at original tempo, at least if there's reason to believe that the instruments were originally tuned to A440. Even if the original tuning isn't known, it might be possible to tell whether the apparent tuning stays the same from the start to the end of a cassette. By the way, on the subject of timing tempos, does anyone know of software that can do a truly, no-nonsense, good job of detecting the beat/tempo of something like typical contra dance music (mixed with crowd noise) with minimal or no hunan assistance? To give you an idea of what I mean by "a truly, no-nonsense, good job", I have a stopwatch designed to time multiple laps of races. Suppose I click the lap/split button on beat 32 of each round of a tune, and suppose that my clicks are usually accurately within 100 milliseconds and very rarely off the beat buy as much as 200 milliseconds. You can do the arithmetic and see that if the tune is played 15 or so times through, I should get a pretty good idea of what the tempo is and of whether the band sped up or slowed down during the course of the tune. I'm asking about software that can generally do at least as well as that, and that, in the cases where it can't track the beat with high confidence, will report that it can't, rather than reporting a bogus tempo. A number of my cassettes were recorded on a recorder that was capable of applying Dolby noise reduction, and I might not always have remembered to check the little boxes on the cassette labels to indicate when cassettes were recorded with noise reduction on. Probably there are DSP techniques to make a good guess about whether a recording was made with Dolby on, but I have no idea what existing software products have this capability. Given a player that supported Dolby, I could of course play any given recording with and without Dolby and decide which way I liked it better, but reliable automatic detection would save me the effort and avoid guessing. Another area where automated DSP might in principle be helpful is in making a preliminary guess at how a recording divides up into segments. Presumably portions with a band playing will have spectral characteristics significantly different from those consisting primarily of crowd noise and/or the caller's voice. Again I know nothing about specific software products (if any) that attempt to segment a long audio file into parts with different spectral characteristics. Here, of course, I'm talking about segments with length typically on the order of a minute to ten minutes, not the vastly shorter segmentation of speech into phonemes. Cassettes that have been stored unplayed for two or three decades might have acquired some crosstalk--faint "echoes" or "pre-echoes" resulting from transfer of magnetization between adjacent layers of tape. I don't know what software, if any, is available to detect and remove/reduce crosstalk without introducing more annoying artifacts than the ones it's removing. That's about what I can think of for now. I'd be interested in information about good products to help with any of the tasks I've described, but the ones I consider most important are * doing the initial capture to WAVE format, * cutting some of the tape hiss and transport noise, and * applying and labelling index marks so that I can jump quickly to whatever segment I want to listen to, without a lot of fast-forwarding, overshooting, rewinding, etc. Thanks for any advice. --Jim