We turned EmbeddingGemma 2 into a speech-to-text model
eg2.haku.shamrrs
A developer created a speech-to-text model by adapting EmbeddingGemma 2, Google's lightweight embedding model, for audio transcription tasks. The implementation is available for experimentation at the provided URL.
Why it mattersThis demonstrates how existing embedding models can be repurposed for new modalities, potentially offering a lightweight alternative to larger speech-to-text systems that readers could deploy locally.
Read at eg2.haku.sh