Microsoft's VibeVoice-ASR-streamingSending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.Full definition →-1.5B lets an application supply names and technical terms to guide speech recognition. These customized hotwords give the recognizer vocabulary relevant to a conversation, which can help with specialized content. Its model card also describes continuous transcription with speaker attribution and support for ten languages.
The release provides an MIT-licensed model and links to implementation code. Its documented contribution is a streaming recognizer that applications can adapt with vocabulary input; the card's text does not quantify how much that input improves a particular user's recordings.
Terms in this piece · Glossary
streaming — Sending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.
