yandex-stt: long recordings to text with SpeechKit
yandex-speechkit-talk-recognition
A Node.js command-line tool that turns long conversation recordings into text through Yandex SpeechKit asynchronous recognition
- 6 ★ on the repository
- Last commit
Extended access
Why this level
- The conversation recording is uploaded to Yandex Object Storage and may contain personal data
- Service account keys are stored in config.js on disk
Install
- Services
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/yandex-speechkit-talk-recognitionDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The conversation recording is uploaded to Yandex Object Storage and may contain personal data
- Service account keys are stored in config.js on disk
This is third-party code. Review the repository files before installing.
What it does
The yandex-stt command takes a file with a voice and runs four steps: converts it to OGG Opus with ffmpeg, uploads it to Yandex Object Storage, sends it to long audio recognition and polls for the result every ten seconds. Data is stored in ~/yandex-stt/. It needs a service account with an API key and a static key with an ID and secret, set in config.js. The package also has a server mode on Express with file upload.
Who it is for. Developers and analysts who transcribe call and meeting recordings and can run a CLI from an agent terminal.
Good fit when
- You have long Russian conversation recordings to turn into text
- Deferred recognition suits you: it is cheaper, about 15 kopecks per minute according to the README
- An agent should start transcription from the terminal and pick up the finished text from the folder
- You want a text file that an agent then analyses or summarises
Not a fit when
- You need real-time recognition: the tool works with files and polls for results
- Recordings must not go to Yandex cloud: the file is uploaded to Object Storage
- You need an MCP server or a skill: the project offers only a CLI and a server mode
Example request
Transcribe the call recording from call.mp3 with yandex-stt and show the textLimitations
The project is not agent-aware: it is a plain CLI tool that an agent has to run by command. It needs ffmpeg 4.x, a Yandex Cloud account and paid recognition. The file is uploaded to Yandex Object Storage. Dependencies in package.json are old, axios 0.21 for example, and the license is ISC. Six stars, one author, last commit in October 2026. The README does not describe the result format or error handling, and its to-do list includes file permissions and editing recognised text, so some features are unfinished. The package also has an Express server mode, but the README does not document it.
How to disable. Remove the package with npm uninstall -g yandex-speechkit-talk-recognition, delete ~/yandex-stt/ and revoke the service account keys.
FAQ
How much does recognition cost?
The README puts deferred recognition at about 15 kopecks per minute, roughly 10 rubles per hour; current prices are on the Yandex Cloud pricing page.
What dependencies are needed?
Node.js, ffmpeg 4.x (3.4 will not work), a Yandex Cloud service account and a static key for Object Storage.
Where is the result saved?
In the ~/yandex-stt/ folder in the user profile.
Is a Yandex Cloud key needed?
Yes, you need a service account with an API key and a static key with an ID and secret for Object Storage.
Which input format is supported?
Anything ffmpeg can read: the tool converts the file to OGG Opus before uploading.
Discussion
No questions or reviews yet.
Related
All Yandex Cloud toolsA skill and scripts for transcribing audio through Yandex SpeechKit: audio preparation, Object Storage upload and asynchronous v3 recognition
A skill with two scripts: text to an .ogg voice message and voice to text through Yandex SpeechKit, with ten voices and three emotions
Four single-file MCP servers for YandexGPT, text recognition, SpeechKit speech synthesis and Yandex search
reliable yandex wordstat collector
Yandex Wordstat Collector
cli collector for yandex wordstat frequencies with failure recovery: null results are never treated as complete data
A set of MCP servers and skills for the OpenClaw assistant: YandexGPT, OCR, speech synthesis and image generation
Yandex Object Storage: MCP, CLI and skill for Yandex Cloud S3
yandex-s3-mcp-ai-cli
One Go binary: an MCP server, CLI and skill for Yandex Cloud S3, read-only by default, with write and delete enabled by flags
README in short
The README briefly describes the four steps, the ffmpeg dependency, running with yandex-stt, the cost of deferred recognition and a to-do list.