Chinar writes punctuation and capitals, so a transcript is ready to put in front of a supervisor, an auditor or a customer. Every model also labels who said what; Chinar-L adds numbers as digits.
A wall of unpunctuated lowercase text is a transcript only in the technical sense: somebody still has to rewrite it before it can be read, quoted or filed. Every Chinar model produces Azerbaijani text with punctuation and capitalisation already in place and who is speaking marked, so the transcript goes straight into the summary, the report or the compliance record it was made for.
Punctuation and capitalisation from all three models, not a raw stream of words.
Speaker labels from every model, so you can see who said what.
Chinar-L also writes numbers as digits, with 93–94% word accuracy on clear Azerbaijani speech, measured against human-checked transcripts.
Fit for documents, summaries and compliance records.
Handles real phone-line audio, not just studio recordings.
Word accuracy for Chinar-L, the GPU model, on clear Azerbaijani speech, measured against human-checked transcripts.
It is trained on genuine call-centre recordings, so ordinary phone-line quality is expected. Noisier audio scores below the clear-speech figure.
When you transcribe every call — QA, search, analytics. It is our most accurate model on real phone calls, and its transcripts are punctuated and speaker-labelled too.
Built for Azerbaijani from the ground up, not adapted from a related language.
Chinar-F is cheap per hour on CPU or GPU — transcribe every call, not a sample.
Global services asked for Azerbaijani give back confident Turkish that looks like a working transcript.
Recordings stay on your infrastructure: no metering, no third-party retention.
See the complete product: problem, features, how it works and deployment.
Request a demo and we will run Chinar against your own Azerbaijani audio — Chinar-F for the volume, Chinar-L for documents, Chinar-P where there is no GPU.