paul-bokelman
commented
3 days ago This will be very difficult and may require that we change our entire analysis approach. Suggestion: we could employ a type of 'incremental analysis' where timestamps are generated in a linear fashion with additional context on each iteration. This type of approach may lead to more cohesive compilation
Description
currently sound effects are trimmed from that start to the end of the interval, instead use gemini to identify the desired 'internal interval' and trim sound effect with respect to that interval in processing
Parent: #8