IBM's Granite Speech 5.0 Adds Speaker Attribution
The Apache-2.0 speech model handles transcription, translation, and figuring out who said what.
IBM has released Granite Speech 5.0, the latest entry in its Granite family aimed at speech understanding. The model is built for three core tasks: transcribing audio, translating it, and attributing speech to individual speakers—all under a permissive Apache-2.0 license.
What it does
According to IBM, the release focuses on three capabilities:
- Transcription — turning spoken audio into text
- Translation — converting speech across languages
- Speaker attribution — identifying who said what in a recording
IBM frames the update around doing all of this "more accurately and faster," positioning speaker attribution as a notable addition. That last feature matters for real-world audio like meetings, interviews, and call-center logs, where knowing the speaker is often as important as the words themselves.
Why it matters
Open, commercially usable speech models remain comparatively scarce next to the flood of open text models. By shipping Granite Speech 5.0 under Apache-2.0, IBM gives developers a foundation they can deploy and fine-tune without licensing friction. The combination of transcription, translation, and attribution in one model is convenient for teams that would otherwise stitch together separate tools.
Full details, including supported languages and performance claims, are available in IBM's announcement on Hugging Face.
Sources
- Visit
Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster
Announcement
More from IBM
All IBM releases →IBM's Granite Switch 4.1 Unifies Chat, Logic, Speech
A single endpoint now routes text conversation, decision-making, and speech synthesis across IBM's Apache-licensed Granite line.
TII extends the Falcon family into speech recognition
The Abu Dhabi lab behind the Falcon language models releases Falcon ASR, an open speech-to-text system.
IBM's Granite 4.2 Adds Reasoning to Open LLM Line
The latest update to IBM's Apache 2.0 model family leans into structured reasoning while keeping its enterprise-friendly licensing.
More in Speech → Text
All Speech → Text →TII Brings Falcon to Speech With New ASR Model
The team behind the Falcon language models expands into speech-to-text with its first automatic speech recognition release.
Cactus Compute's Whistle brings speech-to-text to the edge
A compact on-device ASR model built for edge hardware and WebAssembly, with early support for English, German, and French.
Phonon-2 brings on-device ASR to Apple Silicon
A low-bit quantized, Parakeet-based speech recognizer built to run locally on Mac hardware.
0 comments
No comments yet. Be the first to weigh in.