On Thu, 2012-03-29 at 10:08 -0500, Jeremiah Benham wrote: > Ok. I got this working. I have not tested it much yet though. We will > need to have some visual indicator of the fact that it succeeded in > loading the audio. Yes, there is a lot of work to do on the interface, but it is not worth doing this before establishing whether the idea is workable - is it practical to sync up the audio and score by this method on a reasonable cross section of audio pieces? The second piece I tried I discovered had a change of time signature in it, which it will be possible to adapt the sync script to allow for, but, again, only worth doing if the basic idea is sound. The piece I have done most work on is a piece of old-time fiddle playing, where, although the overall synchronism is kept within the bars you can hear the fiddle notes not landing on the notated timings. Whether this is because the fiddler plays in an uneven manner or because the timings for the start of each measure are inaccurate I'm not sure. The timings for the start of each measure (computed by me beating the start of each measure with the space bar, using the Synchronize to Audio script) are distinctly "noisy", there seems to be an irreducible randomness of a few percent. Again, where this comes from (the amount of the randomness that is) I am not sure. There is scope for smoothing out the randomness of course, I have thought about that and it is not trivial, especially if you allow for both gradual tempo changes and abrupt ones (let alone tempo events taking place mid-bar). I did get a transcription of the fiddle piece, and, in fact, I learned something - the actual notes and rhythms played varied with the repeated strains, and they were not quite the obvious ones; doing the transcription simply by ear I am sure I would have rendered it in a rather conventional version, not capturing the actual variety in there.
What I most would like to know is, how does the synch work on other machines, faster or slower, than mine... Richard _______________________________________________ Denemo-devel mailing list [email protected] https://lists.gnu.org/mailman/listinfo/denemo-devel
