{"skill":{"slug":"deepgram-asr","displayName":"Deepgram Asr","summary":"Transcribe audio via Deepgram Nova-3 API. Fast, accurate, and cost-effective speech-to-text for 50+ languages. Transcripción de audio rápida y precisa con De...","description":"---\nname: deepgram-asr\ndescription: \"Transcribe audio via Deepgram Nova-3 API. Fast, accurate, and cost-effective speech-to-text for 50+ languages. Transcripción de audio rápida y precisa con Deepgram Nova-3. Use when the user needs transcription via Deepgram, or wants fast/cheap cloud transcription.\"\nhomepage: https://developers.deepgram.com/docs/getting-started-with-pre-recorded-audio\nmetadata:\n  {\n    \"openclaw\":\n      {\n        \"emoji\": \"🔊\",\n        \"requires\": { \"bins\": [\"curl\", \"python3\"], \"env\": [\"DEEPGRAM_API_KEY\"] },\n        \"primaryEnv\": \"DEEPGRAM_API_KEY\",\n      },\n  }\n---\n\n# Deepgram ASR (Nova-3)\n\nTranscribe audio files via Deepgram's Nova-3 model. Sub-300ms latency, 50+ languages, $0.0043/min. Free $200 credit on signup.\n\nTranscriba archivos de audio con el modelo Nova-3 de Deepgram. Latencia inferior a 300ms, más de 50 idiomas, $0.0043/min. $200 de crédito gratis al registrarse.\n\n## Sending audio to OpenClaw\n\nCurrently, audio files can be sent to OpenClaw via **Discord** or **WhatsApp**. Send the audio file in a chat message and ask the bot to transcribe it.\n\nActualmente, los archivos de audio se pueden enviar a OpenClaw a través de **Discord** o **WhatsApp**.\n\n> **Note**: Direct voice recording in the OpenClaw web UI is not yet supported. Use a messaging app to send pre-recorded audio files.\n>\n> **Nota**: La grabación de voz directa en la interfaz web de OpenClaw aún no está disponible. Use una aplicación de mensajería para enviar archivos de audio pregrabados.\n\n## Quick start\n\n```bash\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a\n```\n\nDefaults:\n\n- Model: `nova-3`\n- Output: `<input>.txt`\n\n## Useful flags\n\n```bash\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --out /tmp/transcript.txt\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --language es  # Spanish / Español\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --json --out /tmp/result.json\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --speakers  # speaker diarization / separación de hablantes\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --paragraphs  # smart paragraphs / párrafos inteligentes\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --summarize  # AI summary / resumen con IA\n{baseDir}/scripts/transcribe.sh /path/to/audio.m4a --detect-language  # auto-detect language / detección automática de idioma\n```\n\n## How it works\n\nThe script sends audio directly to **Deepgram's API** (`api.deepgram.com`) via HTTPS. No third-party services are involved — audio goes only to Deepgram.\n\n1. Reads the local audio file\n2. POSTs it to `https://api.deepgram.com/v1/listen` with your API key\n3. Parses the JSON response and extracts transcript text\n4. Saves to output file\n\n> **Privacy**: Audio is sent directly to Deepgram's servers over HTTPS. No data is stored by this skill; Deepgram's data retention policy applies.\n\n## Dependencies\n\n- `curl` — for API calls\n- `python3` — for JSON response parsing (stdlib only, no pip packages needed)\n\n## Credentials\n\n1. Sign up at https://console.deepgram.com\n2. Create an API key at https://console.deepgram.com/project/api-keys\n3. Free tier: $200 credit on signup\n\nSet the environment variable:\n\n```bash\nexport DEEPGRAM_API_KEY=\"your_api_key\"\n```\n\n## Supported languages\n\nNova-3 supports 45+ languages. Common language codes:\n\n| Language | Code | Language | Code |\n|----------|------|----------|------|\n| English | `en` | Spanish / Español | `es` |\n| French / Français | `fr` | German / Deutsch | `de` |\n| Portuguese / Português | `pt` | Japanese / 日本語 | `ja` |\n| Chinese / 中文 | `zh` | Korean / 한국어 | `ko` |\n| Hindi / हिन्दी | `hi` | Russian / Русский | `ru` |\n| Italian / Italiano | `it` | Dutch / Nederlands | `nl` |\n| Arabic / العربية | `ar` | Turkish / Türkçe | `tr` |\n\nUse `--detect-language` to auto-detect, or `--language <code>` to specify.\n\nFull list: https://developers.deepgram.com/docs/models-languages-overview\n\n## Supported formats\n\nWAV, MP3, MP4, M4A, OGG, FLAC, WebM, and more.\n","tags":{"latest":"0.3.0"},"stats":{"comments":0,"downloads":245,"installsAllTime":9,"installsCurrent":0,"stars":0,"versions":6},"createdAt":1772007979016,"updatedAt":1778491636632},"latestVersion":{"version":"0.3.0","createdAt":1772033973508,"changelog":"修复shell注入安全漏洞(输入校验+mktemp+curl --)；改为英语+西班牙语双语；删除错误的Telegram描述","license":null},"metadata":{"setup":[{"key":"DEEPGRAM_API_KEY","required":true}],"os":null,"systems":null},"owner":{"handle":"vahnxu","userId":"s17d86qgf1fm0mqac9pvp4rrns83tdzh","displayName":"vahnxu","image":"https://avatars.githubusercontent.com/u/19759725?v=4"},"moderation":null}