Uzbek speech, right inside your AI agent
The clarion-mcp server adds Uzbek speech recognition to Claude Code, Codex and Cursor: the agent turns audio and video into punctuated text, timestamped segments or SRT and VTT subtitles.
>
Why it helps
The agent takes care of everything around a recording, from upload to the finished file.
Subtitles for video
Ready-made SRT and VTT for YouTube, lessons and social media. Frames are already split by length and timing.
Transcripts and summaries
Interviews, calls, lectures: the agent gets the text and immediately writes notes, pulls out key points or translates.
Whole folders at once
An archive of podcasts or call recordings in one sentence. The agent walks through the files and saves results next to them.
No code, no keys
No scripts against the API and no keys to store. Allow access in the browser once and keep working in a normal conversation.
How it works
Connect once, then just talk to your agent.
- 1
Connect
Add Clarion to your agent with one command and allow access in the browser.
- 2
Ask
Say what to do with the recording: transcribe it, make subtitles, sum it up.
- 3
Get the result
The agent uploads the file, waits for recognition and saves the text or subtitles.
Connect
One command and a confirmation in the browser. No key needed.
claude mcp add --transport http clarion-mcp https://api.uzasr.uz/mcpcodex mcp add clarion-mcp --url https://api.uzasr.uz/mcpcodex mcp login clarion-mcp/plugin marketplace add rustammdev/clarion-mcp/plugin install clarion-mcp@clarion-mcpAfter the command, open /mcp in Claude Code, pick clarion-mcp → Authenticate and click “Allow” in the browser. In Codex, codex mcp login clarion-mcp starts the sign-in, and the plugin opens the browser on its first call. Connected agents are listed in your dashboard under “API keys”.
What you can ask
Copy an example and paste it into your agent.
«Make Uzbek subtitles for dars.mp4 and save them to dars.srt»
«Transcribe intervyu.mp3 and briefly list the key points»
«Transcribe every mp3 in the podcasts/ folder»
What you get
The agent picks the format that fits your task.
- Text
- Continuous text with punctuation, in Uzbek Latin script.
- Timestamped text
- Every phrase with its start and end time.
- SRT and VTT subtitles
- Frames up to 6 seconds and 84 characters, built from word timestamps.
Files up to 256 MB and up to 3 hours of audio. For a large video the agent first compresses it to audio without losing speech.
Privacy
What we keep and where the result goes.
- Audio
- Kept on the server briefly and deleted right after recognition.
- Transcript
- Deleted after 7 days.
- Result
- Goes to the agent as a tool response, which means to the agent's LLM provider too. That choice stays yours.
CI and headless servers
Where a browser can't open, connect with an API key.
Sign up, create a key in your dashboard and send it in the Authorization header. For Codex the lines go into ~/.codex/config.toml. Create a separate key per machine so a leak only means revoking that one.
claude mcp add --transport http clarion-mcp https://api.uzasr.uz/mcp \ --header "Authorization: Bearer $UZASR_API_KEY"[mcp_servers.clarion-mcp]url = "https://api.uzasr.uz/mcp"bearer_token_env_var = "UZASR_API_KEY"