name: hermes-soniox repo: https://github.com/XimilalaXiang/hermes-soniox sha: 79f8d1dec5fcdc75e22ec34e857e5b4e2b8828bf description: >- Native BYOK Soniox text-to-speech and asynchronous speech-to-text providers with configurable languages and voices, automatic STT language detection, and Hermes voice-bubble support. Disclosure — TTS sends reply text and synthesis settings to tts-rt.soniox.com; STT uploads audio, optional language hints, and optional prompt context to api.soniox.com, then best-effort deletes the remote transcription and uploaded file. SONIOX_API_KEY is resolved through Hermes' active-profile secret scope. No telemetry, shell commands, background processes, or stored credentials. maintainer: XimilalaXiang tier: community version: "0.2.1" category: voice requires_hermes: ">=0.22.4" docs_url: https://github.com/XimilalaXiang/hermes-soniox#readme platforms: [] known_issues: - Realtime WebSocket TTS/STT is not implemented; v0.2.1 uses file-based synthesis and asynchronous transcription. capabilities: provides_tools: [] provides_hooks: [] provides_middleware: [] requires_env: - SONIOX_API_KEY