{
  "markdown": "# claude-telegram-voice\n\nClaude Code plugin that automatically transcribes Telegram voice messages so Claude can understand what you said.\n\nWorks with any transcription tool — [OpenAI Whisper](https://github.com/openai/whisper), [faster-whisper](https://github.com/SYSTRAN/faster-whisper), [whisper.cpp](https://github.com/ggerganov/whisper.cpp), OpenAI API, or any custom script.\n\n## How it works\n\n1. You send a voice message to your Telegram bot\n2. The Telegram plugin delivers the message to Claude with a file reference\n3. This plugin's skill tells Claude to automatically download and transcribe it\n4. Claude receives the spoken text and responds normally\n\nNo modifications to the official Telegram plugin required.\n\n## Prerequisites\n\n- [Claude Code](https://docs.anthropic.com/en/docs/claude-code) with the [Telegram plugin](https://github.com/anthropics/claude-plugins-official) installed\n- A transcription tool of your choice (see examples below)\n\n## Quick start (4 steps)\n\n```\n/plugin marketplace add Undermove/claude-telegram-voice\n/plugin install claude-telegram-voice\n/claude-telegram-voice:setup\n```\n\nThen restart Claude Code — done! The setup command will check your system, help install a transcription tool if needed, and configure everything automatically.\n\n## Alternative installation\n\n### From official marketplace (when available)\n\n```\n/plugin install claude-telegram-voice\n```\n\n### Local / development\n\n```bash\ngit clone https://github.com/Undermove/claude-telegram-voice.git\nclaude --plugin-dir ./claude-telegram-voice --channels plugin:telegram@claude-plugins-official\n```\n\n### Manual setup\n\n#### 1. Create a transcription script\n\nThe plugin calls your script with one argument — the path to an `.ogg` audio file. Your script must print the transcription to stdout.\n\n**Example: Local Whisper**\n\n```bash\n#!/bin/bash\n# transcribe.sh\nwhisper \"$1\" --model small --language ru --output_format txt --output_dir /tmp 2>/dev/null\ncat \"/tmp/$(basename \"${1%.*}\").txt\"\nrm -f \"/tmp/$(basename \"${1%.*}\").txt\"\n```\n\nInstall Whisper: `pip install openai-whisper`\n\n**Example: faster-whisper**\n\n```bash\n#!/bin/bash\n# transcribe.sh\nfaster-whisper \"$1\" --model small --language ru 2>/dev/null | grep -v '^\\['\n```\n\nInstall: `pip install faster-whisper`\n\n**Example: OpenAI API**\n\n```bash\n#!/bin/bash\n# transcribe.sh\ncurl -s https://api.openai.com/v1/audio/transcriptions \\\n  -H \"Authorization: Bearer $OPENAI_API_KEY\" \\\n  -F file=\"@$1\" -F model=\"whisper-1\" | jq -r .text\n```\n\n#### 2. Make the script executable\n\n```bash\nchmod +x /path/to/transcribe.sh\n```\n\n#### 3. Configure the environment variable\n\nAdd to `~/.claude/channels/telegram/.env`:\n\n```\nVOICE_TRANSCRIBE_CMD=/path/to/transcribe.sh\n```\n\nOr set it in your shell before launching Claude Code.\n\n#### 4. Launch Claude Code\n\n```bash\nclaude --channels plugin:telegram@claude-plugins-official\n```\n\nSend a voice message — Claude will automatically transcribe and respond to it.\n\n## Configuration\n\n| Variable | Required | Description |\n| --- | --- | --- |\n| `VOICE_TRANSCRIBE_CMD` | Yes | Path to a script that accepts an audio file path and prints transcription to stdout |\n\nAny tool works — as long as it reads the audio file and prints text to stdout. If the script fails, Claude will let the user know and ask them to resend as text.\n\n## Plugin structure\n\n```\nclaude-telegram-voice/\n├── .claude-plugin/\n│   └── plugin.json          # Plugin manifest\n│   └── marketplace.json     # Marketplace config\n├── commands/\n│   └── setup.md             # /claude-telegram-voice:setup command\n├── skills/\n│   └── voice-transcription/\n│       └── SKILL.md         # Auto-trigger skill\n├── README.md\n└── LICENSE\n```\n\n## License\n\nMIT\n",
  "bytes": 3679,
  "sha": "3a3c139c1ded64c0038e56d2657dc79ced853719ba06ced56762ae733c8b1af6",
  "repo_slug": "undermove/claude-telegram-voice",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_undermove_claude_telegram_voice_claude_t_344bd5b3/readme"
}