An open-weight speech-to-text model for fast local transcription.
F16 default
tokens
Canonical metadata is separated from immutable version and Run evidence.
Echo Transcribe 3B is a demonstration model for non-LLM modalities, latency Runs and language filters.
1 published Runs; each result links to its exact conditions and evidence level.
No comparable speed metrics yet.
Benchmark claims remain tied to immutable revisions even after the repository changes.
Faster timestamp decoding and improved noisy-audio performance.
