● Covered by 1 source · 2 reportsMedium impact2 neutral
IBM Updates Granite Speech Models with Improved Accuracy, Multilingual Support, and Speed
Key points
- granite-speech-5.0-2b offers improved English ASR and multilingual transcription/translation.
- Turbo CTC model now supports seven languages and includes native punctuation/capitalization.
- Models are released under Apache 2.0 and support various inference frameworks.
- granite-speech-5.0-2b ranks second in English WER among open-weight ASR models.
- IBM released Granite Switch 4.1.1, a patch update to its modular language model architecture.
- Granite Switch 4.1.1 includes an updated multilingual speech-to-text ASR model across 10 languages.
- Granite Switch 4.1.1 includes an improved Decision adapter in granitelib-core.
- The Decision adapter shows improved performance on the Jev Decision Index test set.
New Granite Speech Models Released
IBM has announced three updates to its Granite Speech collection. The new speech-language model, granite-speech-5.0-2b, is introduced alongside revised versions of granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus. These models are designed to cover enterprise speech tasks including transcription, translation, fast edge transcription, and speaker attribution.
granite-speech-5.0-2b Enhancements
The granite-speech-5.0-2b model, a successor to granite-speech-4.1-2b, maintains the same parameter count. It transcribes English, French, German, Spanish, Portuguese, and Japanese, and translates to and from English in these languages, plus English to Italian and Mandarin. Training involved 390,000 hours of public audio and synthetic data for Japanese ASR, keyword biasing, and translation.
Compared to its predecessor, granite-speech-5.0-2b offers more accurate English ASR, better handling of noise and distant-microphone speech, and language ID prompting for multilingual transcription. It also features improved long-form and conversational translation, support for keyword lists up to 2,000 terms, and on-request punctuation and capitalization.
Turbo CTC Model Updates
The granite-speech-5.0-470m-turboctc model, originally an English-only encoder focused on speed, now includes multilingual transcription for English, Japanese, Spanish, German, French, Italian, and Portuguese. It also provides native punctuation and capitalization in transcripts and is more robust in difficult acoustic environments due to targeted training on reverberant audio.
Performance and Availability
granite-speech-5.0-2b ranks second among open-weight ASR models with an average English Word Error Rate (WER) of 5.06% on the Open ASR Leaderboard's public English short-form test sets. All models are released under the Apache 2.0 license and are supported in Transformers, vLLM, llama.cpp, and mlx-audio for Apple Silicon.
Updates
- IBM released Granite Switch 4.1.1, a patch update to its modular language model architecture.
- Granite Switch 4.1.1 includes an updated multilingual speech-to-text ASR model across 10 languages.
- Granite Switch 4.1.1 includes an improved Decision adapter in granitelib-core.
- The Decision adapter shows improved performance on the Jev Decision Index test set.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
How outlets covered it
IBM released Granite Switch 4.1.1, a patch update to its modular language model architecture. This version includes an updated multilingual speech-to-text ASR model and an improved Decision adapter, demonstrating the platform's ability to update components incrementally.
IBM released updates to its Granite Speech collection, introducing granite-speech-5.0-2b and revised versions of granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus. These updates provide more accurate transcription and translation across multiple languages, faster edge transcription, and improved speaker diarization capabilities for enterprise speech applications.