Tags: OpenMind/OM1
Tags
Fix robot name (#2478) * Switch TTS provider to ElevenLabs and set voice_id Update config/greeting.json5 to use ElevenLabs for TTS: replace kokoro tts_provider with "elevenlabs" across hooks, add a voice_id (PoHUWWWMHFrA8z7Q88pu) to the connector config and handler configs, and remove kokoro-specific model_id and base_url entries. This switches the greeting flow to the ElevenLabs voice backend. * Configurable ElevenLabs voice & Gemini model Replace hardcoded ElevenLabs voice_id in config/greeting.json5 with environment variable interpolation (${VOICE_ID:-<default>}) so the TTS voice can be overridden at runtime while preserving the existing default. Update the default Gemini model in src/llm/plugins/gemini_llm.py from GEMINI_3_FLASH_PREVIEW to GEMINI_3_1_FLASH_LITE_PREVIEW to use the newer/lighter model. * Fix robot name
Switch to GeminiLLM and use ROBOT_NAME env (#2469) Update greeting configs to make the agent name configurable via ${ROBOT_NAME:-Bits} in both config/greeting.json5 and config/greeting_local.json5. Change cortex_llm.type from OpenAILLM to GeminiLLM in the main config and remove the hardcoded model setting there. Local config retains QwenLLM and its model/base_url but now uses the environment-backed agent_name. History lengths left unchanged.
test: add unit tests for 7 remaining action connectors (#2227) Add tests for face/ros2, speak/ros2, speak/zenoh, speak/ub_tts, move_go2_autonomy/idle, greeting_conversation/elevenlabs and kokoro. All 7 connector source files now have 100% test coverage (51 tests). Also add missing __init__.py files for speak/ and move_go2_autonomy/. Co-authored-by: openminddev <147775420+openminddev@users.noreply.github.com>
Rewrite core endpoints docs (#2111) * Revise and expand API reference docs Expanded and clarified multiple API reference pages: updated pricing copy (rates, memory notes, shipping timelines, and plan features) and harmonized changes across mintlify copies. Significantly enhanced endpoint docs: added base URLs, authentication details (JWT/Clerk for core account endpoints, x-api-key for ElevenLabs, API key query for Google ASR), comprehensive request/response examples, error cases, and usage notes for account & key management, ElevenLabs TTS, and Google ASR (rewritten as a WebSocket streaming API with Python/Node/wscat examples and best practices). Renamed and enriched the LLM chat completions doc with supported providers/models, request/response schemas, examples, and advanced parameters. Overall: large content additions, examples, and consistency fixes across docs/api-reference and mintlify/api-reference. * Resolve merge conflict and update docs Clean up documentation: remove leftover merge conflict markers and an outdated curl example in the LLM endpoints file; update a note link label in the Riva endpoint docs from "GitHub Repo" to "Our GitHub"; and correct multiple pricing entries in the introduction pricing table (several OpenAI, GPT-4.1/4.1-mini/nano, GPT-5 variants, DeepSeek, Gemini, grok, qwen, and llama entries) to their accurate values. * Standardize GitHub link text in docs Update Note blocks in riva and vila_vlm endpoint docs (and their mintlify copies) to use consistent link text 'Our GitHub' instead of 'the Our GitHub' or 'GitHub Repo'. This normalizes wording across the documentation.
Fix LLM chat history (#844) * Fix chat history message type * Enhance Input dataclass with tick counter and update IOProvider to track ticks in inputs * Add tests for input tick filtering in LLMHistoryManager Introduces comprehensive tests to verify that LLMHistoryManager only adds inputs from the current tick to the history. Tests cover scenarios with multiple ticks, no inputs for the current tick, and input updates at tick boundaries. Also includes minor cleanup in IOProvider and its tests.