Diarization usually means a gated, token-walled model. For the common two-speaker call, a few dozen lines of pitch clustering beat reaching for the big model. A story about choosing the simpler tool.
In the first post I argued there are two ways to pull meaning out of audio:
measure it with signal processing, or estimate it with a model. This post
is the story of a problem where the obvious move was to estimate — and where
measuring turned out to be better.
The problem: labeling who is speaking. A transcript that says "Agent: …" and
DAX, the audio SSP and sales company, is working to prove that fragmented audio inventory can deliver scale and outcomes for…
A complete walkthrough of speech transcription, LLM inference, tokenization, and 4-bit quantization....
Binary measurement preserves margin and lets late-arriving impressions claim credit they may not deserve. We need a new…
Thanks to Dipankar Sarkar, Mike Czerwinski, Max Quimby, and Ponsubash Raj R for the detailed comments...
There's a school of voice-agent development that treats turn detection as something you buy, not...
Four provenance-tracking arms, identical gates, one uncompacted oracle: measuring exactly what memory compaction does to agent…
