EmbeddingGemma 2 used for speech-to-text: 1 hour of audio transcribed locally in ~13 seconds
GlennCameronjr · x · 2026-10-23
An experiment from the Google Gemma team skips traditional ASR entirely, instead using EmbeddingGemma 2's unified audio-text vector space to match speech against text via vector similarity.
The result: transcribing an hour of audio in about 13 seconds, fully locally — a creative demonstration of cross-modal embedding alignment for speech.
More from Models
- Sophont's Holon-S1 predicts genes and biomarkers from routine pathology slides — iScienceLuvr · 2026-10-23
- Microsoft says its MAI-DxO AI diagnoses hard cases 83% vs doctors' 36% — RebeccaBellan · 2026-10-23
- Doc parsing lite mode scores 75.1 vs Azure Read's 42.3 on OmniParseBench, dominates non-Latin scripts — VikParuchuri · 2026-10-23
- Datalab open-sources Indra OCR: a 1M-parameter model that beats PaddleOCR on CPU — VikParuchuri · 2026-10-23
- The inside story behind OpenAI's million-dollar math proof — scientificamerican · 2026-10-23
- Anthropic's anti-distillation defense cuts attacks 90-99%, but expert data is still for sale — johnkorrigan · 2026-10-23