search

patrick99e99 / BlueWizard

LPC analysis tool for the Texas Instruments TMS5220 chip
http://atariage.com/forums/topic/260857-introducing-bluewizard-qboxpro-replacement-speech-analysis-tms5220-tool/
MIT License
49 stars 16 forks source link

repeat flag #15

Open deladriere opened 6 years ago

deladriere commented 6 years ago

analyzing sounds from some old TMS5220 archive shows lots of repeats. I like them because my instruments use them a looping point

I noticed that the repeat flag is never generated when I process sound with BlueWizard Is it on purpose or it's just not yet implemented?

Lord-Nightmare commented 4 years ago

I believe it just wasn't implemented. A naive implementation would be to issue a repeat flag if all the coefficients except pitch and energy changed. A much more complex implementation would allow setting a threshold for the total amount of change, weighted by the coefficient number (the lower coefficients are much more sensitive), and to 'throw out' certain data (for the new frame? or for the older one?) if it is almost but not quite identical to the previous frame. This has to be done carefully so that the Lattice Filter doesn't oscillate when modifying the data for older frames, and has to be done carefully so that if a parameter was slowly sweeping upward or downward it can't get too over-quantized using repeat frames.