> On 22:35 Uhr David W. Fenton wrote: > > Well, it depends on how it's doing it. The only thing I could figure > > out how to do on the software I downloaded was real-time analysis of > > something playing on my PC. This, of necessity, would included the > > basic noise and distortion built into my PC's audio system. > > No it wouldn't. It's analysing the digital code being streamed from the > soundfile. Unless it is doing something very unusual there is no way the > audio system inside your PC (which does everything from DA conversion > onwards, but nothing before it) could have any influence on the analysis. > > How else would you analyse a sound file than in realtime? You cannot > make a frequency analyis without playing the file, since there are no > frequencies actually recorded in the file, all that is recorded is many, > many, many snapshots of the soundwave. Any of these snapshots only gives > the coordinates for a single point of the curve. > > Frequencies only come into play when the soundfile is played back or > otherwise analysed with respect to time. > > Whatever the software is doing it is definitely dealing with the digital > data, not with any analogue translation of it. > > Johannes
You are right that the software of necessity must analyze the digital stream before the D/A conversion, Johannes. However, real-time playback is not necessary to do a spectrum analysis. The sampled waveform is used as input to an algorithm known as a Discrete Fourier Transform, or DFT (the digital equivalent to an analog Fourier Transform), and what comes out of the DFT are coefficients of the amplitude of the frequency spectrum at an interval determined by the sample rate frequency. This can be easily done in real-time nowadays by any desktop PC, but when I was working in this field in the mid-70's it required extremely fast, specially built hardware (running at all of 10 MHz!) and very clever programming techniques to reduce the calculation overhead. These software techniques are collectively known as the Fast Fourier Transform, or FFT. I think the reason that PC software would do the DFT in real-time instead of directly on the data in the file is to avoid having to write a parser for different audio formats, or even for different flavors of the same format. If the samples are intercepted when they're sent to the sound card, it doesn't matter what the original audio format was. Lee Actor Composer-in-Residence and Assistant Conductor, Palo Alto Philharmonic http://www.leeactor.com _______________________________________________ Finale mailing list [email protected] http://lists.shsu.edu/mailman/listinfo/finale
