Blog
Notes on speech, transcription and subtitles. Plain, with examples, no hype.
-
From recording to protocol: meeting transcription with a summary API
Upload a meeting recording, poll, fetch transcript and summary with decisions and open points, translate, handle errors and understand billing.
-
Speaker diarization explained: who said what, and where it goes wrong
How speaker diarization assigns speech to S1, S2 and so on, where it fails (overlaps, interjections, similar voices) and how to map names afterwards.
-
SRT vs WebVTT: which subtitle format to use, and how to build both
SRT and WebVTT compared: syntax, where each one is accepted, the 42-character two-line rule, timing limits and how to generate both from word timestamps.
-
Custom vocabulary for speech recognition: names and jargon spelled right
Why speech recognition misspells names and jargon, what context biasing does, glossary vs per-request vocabulary, the limits, and how to measure the effect.
-
Transcribing sensitive recordings under the GDPR: what EU-only really means
What EU processing, zero data retention and delete-after-fetch mean in practice, what a DPA covers, what to ask a transcription vendor, and a checklist.