Chinar-F is a compact model trained on real Azerbaijani calls and cheap per hour on CPU or GPU, so full coverage stops being a budget question: every call transcribed, an archive made searchable, analytics and quality monitoring fed.
Sampling exists because transcription is expensive, not because a sample is what anyone wanted. Chinar-F is a compact model trained on real Azerbaijani call-centre recordings at phone quality, and it runs on CPU or GPU at a low cost per hour — which is what makes transcribing everything realistic rather than aspirational. None of it bills per minute, so the cost does not grow with call volume.
23% fewer word errors than a commercial cloud speech service on the same real calls.
Our most accurate model on real phone calls.
Runs on CPU or GPU, on your own infrastructure.
No per-minute billing: the cost does not grow with call volume.
Four to seven times faster than the cloud services benchmarked against it.
Not on phone calls — Chinar-F is our most accurate model on real calls. Chinar-L is for transcripts people read: it adds punctuation, capitals and numbers as digits, and the 93–94% clear-speech figure belongs to it.
Yes — everything can go through Chinar-F while the calls that end up in a document go through Chinar-L.
Chinar-F runs on CPU or GPU. If the site has no GPU at all, Chinar-P runs 26× faster than real time on four CPU cores.
Built for Azerbaijani from the ground up, not adapted from a related language.
Chinar-L returns punctuated, capitalised Azerbaijani, ready for a document or a compliance record.
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 the transcript people read, Chinar-P where there is no GPU.