speaches releases

v0.9.0-rc.3 New feature · 5mo

⚠ Upgrade required

• The `onnx-export` package has been removed; users relying on it should plan accordingly.
• `huggingface_hub.list_models` deprecated parameters have been addressed — verify any custom integrations using this API.
• Models are now downloaded automatically on startup; ensure sufficient disk space and network access in deployment environments.

Notable features

• Speaker diarization
• OpenTelemetry support
• CLI command aliases

View release on GitHub

v0.9.0-rc.2 New feature · 6mo

⚠ Upgrade required

• This is a release candidate (rc.2); users should treat it as pre-release and validate before production use.

View release on GitHub

v0.8.3 Bug fix · 8mo

Fixed corrupted audio headers for WAV and FLAC streaming, improved Piper model validation, and added CA certificates to Docker image.

View release on GitHub

v0.8.2 New feature · 8mo

⚠ Upgrade required

• VAD filter is now enabled by default for STT endpoints. Users who relied on previous behavior (VAD filter off) should explicitly set VAD_FILTER=False in their configuration to preserve existing transcription behavior.

Notable features

• VAD filter enabled by default (configurable via VAD_FILTER config option)
• Major Apple Silicon ONNX speedup with actionable error handling and guidance
• TTS performance benchmarking script added

View release on GitHub

v0.8.1 Bug fix · 8mo

Fixed Piper model ID parsing issues and legacy library metadata handling.

View release on GitHub

v0.8.0 Breaking risk · 8mo

⚠ Upgrade required

• Remove `default_response_format` and `default_language` from your configuration before upgrading (#313)
• Live transcription is deprecated; migrate to the new real-time API

Breaking changes

• Removed `default_response_format` configuration option (#313)
• Removed `default_language` configuration option (#313)
• Live transcription deprecated

Notable features

• Real-time API
• Real-time WebRTC support
• Real-time UI

View release on GitHub