Azerbaijani speech recognition

Most speech engines don’t speak Azerbaijani. They guess.

Ask a global speech service to transcribe an Azerbaijani call and you usually get back fluent, confident text — in Turkish. Chinar is built for Azerbaijani from the ground up, in two sizes: one for transcripts people read, one for transcripts machines read.

The problem

Why Chinar exists

In Turkish
Confident, and wrong

Ask a global speech service for an Azerbaijani transcript and it returns fluent Turkish. It looks like a working transcript — and unless someone on your team reads Azerbaijani, nobody notices until the text reaches a report.

Unsupported
A gap, not a tuning problem

Two major 2026 speech services tested head to head on the same audio — one commercial, one open — produced Azerbaijani output that was effectively unusable. They do not support the language at all.

Read speech
Trained on the wrong audio

Many engines learn from clean, read-aloud recordings and fall apart on a real conversation: background noise, interruptions, two people talking at once, ordinary phone-line quality.

Per hour
Metered, and off your premises

Transcribing every call through a cloud service means paying by the hour and handing customer conversations to a third party that keeps its own copy.

What it does

Built to do the job, end to end

🌳

Azerbaijani from the ground up

Not a multilingual model with Azerbaijani bolted on, and not Turkish with an accent. The language is the design — which is the difference between a transcript you can act on and one that merely looks finished.

📝

Chinar-L, for transcripts people read

Punctuated and capitalised, ready to put in front of a supervisor, an auditor or a customer. Built for call recordings, interviews and meetings where the words end up in a document, a summary or a compliance record.

⚙️

Chinar-F, for transcripts machines read

The same language expertise, roughly fifty times smaller. Built for volume: transcribing every call rather than a sample, powering search across an archive, or feeding analytics and quality monitoring.

🎯

87% word accuracy

Chinar-L scores 87% word accuracy on clear Azerbaijani speech, measured against reference transcripts from a commercial speech system.

📞

Real phone calls, not studio audio

Trained on genuine call-centre recordings — background noise, interruptions, people talking over each other, ordinary phone-line quality.

🌐

Try it before you talk to us

A free hosted version is open at chinar.allmaz.az: upload a recording, read the Azerbaijani transcript, edit and export it. No card, no sales call — judge the language for yourself first.

🏠

Your recordings stay yours

Chinar runs on your own infrastructure. Customer conversations never leave your building: no per-hour metering, no third-party retention, no questions about where the audio went.

Explore in depth

Chinar capabilities

All Chinar capabilities →

How it works

From input to outcome

1Choose the size — Chinar-L for transcripts people read, Chinar-F for volume
2Deploy it on your own infrastructure, inside your own network
3Send audio through the API: call recordings, interviews, meetings
4Chinar-L returns punctuated, capitalised Azerbaijani text; Chinar-F returns the same language expertise at scale
5Feed the transcripts into reports, search, analytics or quality monitoring
Who it's for

Made for the people who use it

Call centers & customer service

Transcribe every conversation instead of a sample, and read the result in Azerbaijani without translating it first.

Compliance & quality teams

Punctuated transcripts fit to sit in a compliance record, produced without sending the recording to a third party.

Analytics & data teams

Run Chinar-F across an archive to power search, analytics and quality monitoring at a cost per hour that makes full coverage affordable.

Media, research & public sector

Interviews and meetings transcribed in Azerbaijani, on infrastructure you control.

Why Chinar

What sets it apart

It actually speaks the language

In head-to-head testing on the same audio, two major 2026 speech services — one commercial, one open — produced Azerbaijani output that was effectively unusable. Chinar handled the same recordings correctly. That gap is not a tuning difference: those services do not support the language at all.

Real phone calls, not studio audio

Chinar is trained on genuine call-centre recordings, with the noise, the interruptions and the phone-line quality that come with them. Many engines are trained on read speech and fall apart on a real conversation.

Four to seven times faster

Measured against the cloud services we benchmarked, so a day of recordings is processed in minutes rather than left in a queue.

Two sizes, one command of the language

Chinar-L and Chinar-F share the same Azerbaijani expertise, so the transcript a person reads and the transcript a machine reads agree with each other — and Chinar-F runs comfortably where a large model will not.

Chinar
chi-NAR
çinar
The name

Where the name comes from

The çinar is the oriental plane tree — the broad, long-lived tree that shades courtyards and village squares across Azerbaijan, old enough that people meet under the same one their grandparents did. It is a local thing, deeply rooted, and it outlasts whatever is planted beside it. The name fits a speech model built for this language rather than borrowed from the one next door.

Deployment & trust

Yours to control

Chinar runs on your own infrastructure, so customer conversations never leave your building: no audio goes to a third party, nothing is retained outside your network, and no per-hour meter runs against a cloud account. The 87% figure is word accuracy for Chinar-L on clear Azerbaijani speech, measured against reference transcripts from a commercial speech system — noisier audio scores lower, and we would rather state the condition than quote a number without it.

Learn more

Chinar — topics, use cases & comparisons

Backed by Smart Solutions

Built by the company that runs national infrastructure.

Allmaz is the AI product studio of Smart Solutions, which built and operates Azerbaijan's unified public procurement portal — established by presidential decree and delivered as one of the country's first public-private partnerships in digital government.

9,000+companies use Smart Solutions products
15+years of national-scale delivery, since 2006
150+people across the group
See the track record →
Get started

See Chinar on your own data

Request a demo and we will run Chinar on your own Azerbaijani recordings — Chinar-L for the transcript people read, Chinar-F for the volume.