[devword]

Curious how you handle long-context transcripts locally. Are you chunking + retrieval, or relying on large-context hosted models through BYO APIs?

[tantara]

STT models I used are fast enough to transcribe 1hr audio on my macbook (m1max).