This plugin has no README in the catalog yet.
sarvam
Sarvam AI for Hermes: speech-to-text (voice notes and live transcription in voice mode) and text-to-speech (Bulbul voices) for 22 Indian languages, plus six tools: sarvam_translate, sarvam_transliterate, sarvam_detect_language, sarvam_read_document (PDF/image OCR), sarvam_translate_document and sarvam_transcribe_recording (long audio with speaker labels). Not affiliated with Sarvam AI. Disclosure — registers one transcription provider, one TTS provider and six tools (toolset "sarvam"), no hooks; requests go to https://api.sarvam.ai (and wss://api.sarvam.ai for live transcription) with SARVAM_API_KEY read through Hermes's secret store; document translation and long recordings upload the file to, and download results from, the signed https storage URL Sarvam returns for that job; sends audio Hermes asks to transcribe, text to speak, tool text, and files whose path the agent supplies; writes the requested audio file and translated document copies beside the original under a new name; one WebSocket only while live transcription runs; no stored state, shell commands, background processes or telemetry; every call is billed by Sarvam to the user's key.
About
Sarvam AI for Hermes: speech-to-text (voice notes and live transcription in voice mode) and text-to-speech (Bulbul voices) for 22 Indian languages, plus six tools: sarvam_translate, sarvam_transliterate, sarvam_detect_language, sarvam_read_document (PDF/image OCR), sarvam_translate_document and sarvam_transcribe_recording (long audio with speaker labels). Not affiliated with Sarvam AI. Disclosure — registers one transcription provider, one TTS provider and six tools (toolset "sarvam"), no hooks; requests go to https://api.sarvam.ai (and wss://api.sarvam.ai for live transcription) with SARVAM_API_KEY read through Hermes's secret store; document translation and long recordings upload the file to, and download results from, the signed https storage URL Sarvam returns for that job; sends audio Hermes asks to transcribe, text to speak, tool text, and files whose path the agent supplies; writes the requested audio file and translated document copies beside the original under a new name; one WebSocket only while live transcription runs; no stored state, shell commands, background processes or telemetry; every call is billed by Sarvam to the user's key.
More by Subramanyarao11
2 plugins by Subramanyarao11 in the catalog.
Developer Mode
Explore built-in Hermes toolsets and actions.

Subramanyarao11