IBM Granite Speech 5.0: Faster, More Accurate Speech Transcription and Translation

IBM has unveiled Granite Speech 5.0, a suite of speech-language models designed to enhance transcription, translation, and speaker attribution. The new models include granite-speech-5.0-2b, granite-speech-5.0-470m-turboctc, and granite-speech-4.1-2b-plus, each tailored for specific enterprise needs.

Granite Speech 5.0 2B: More Accurate and Multilingual

Granite Speech 5.0 2B is the successor to granite-speech-4.1-2b at the same parameter count. It transcribes English, French, German, Spanish, Portuguese, and Japanese, and translates to and from English in those languages, plus English to Italian and Mandarin. The model was trained on 390,000 hours of public audio and synthetic data, focusing on Japanese ASR, keyword biasing, and translation. It offers higher accuracy on English short-form benchmarks and better handling of noise, reverberation, and distant-microphone speech. Language ID prompting lets users name the input language to steer multilingual transcription. Keyword lists of up to 2,000 terms bias transcription and translation toward product names, people, and acronyms in every supported language. Punctuation and capitalization are available on request, with capitalization F1 scores ranging from 89.7 to 99.5 across test sets. The model ships under Apache 2.0 with support in Transformers, vLLM, llama.cpp, and mlx-audio for Apple Silicon.

Granite Speech 5.0 Turbo CTC: Now Multilingual and 1.6x Faster

Granite Speech 5.0 Turbo CTC, previously English-only, now supports multilingual transcription in English, Japanese, Spanish, German, French, Italian, and Portuguese. It also includes native punctuation and capitalization, eliminating the need for a separate truecasing step. Training that explicitly targets reverberant audio makes it more robust in echoey rooms and far-field recordings. By compiling a single Conformer block and reusing its graph across the stack, IBM cut warmup time and pushed Turbo CTC to 19,870 RTFx on the Open ASR Leaderboard, up from roughly 12,600 RTFx at launch — about 1.6x faster at the same accuracy level.

Granite Speech 4.1 2B Plus: Better Speaker Attribution and Timestamps

The updated granite-speech-4.1-2b-plus is tuned for meetings, support calls, and interviews with several speakers. It keeps the same architecture, prompt format, and API, making it a drop-in replacement with improved weights. Speaker attribution improves by 7%, with word diarization error rate (WDER) and speaker-attributed WER (cpWER) both dropping 7%. Average timestamp shift (AAS) is cut by more than half, yielding tighter timing in captions, search, and review. WER on long-form test sets drops 24%, and errors on biased keywords fall 45% with 20-term keyword lists.

Benchmarks and Performance

Granite Speech 5.0 2B ranks #2 among top open-weight ASR models with an average English Word Error Rate (WER) of 5.06%, based on official results as of October 21, 2026, averaged over the Open ASR Leaderboard’s public English short-form test sets. Granite Speech 5.0 Turbo CTC achieves 19,870 RTFx on the same leaderboard, making it significantly faster while maintaining accuracy.

Which Model to Choose

For high transcription accuracy, translation, or keyword biasing, use granite-speech-5.0-2b. For raw speed or on-device transcription, use granite-speech-5.0-470m-turboctc, and try it in the browser with the WebGPU streaming demo. For meetings or multi-speaker calls, use granite-speech-4.1-2b-plus.

All models are released under an Apache 2.0 license with native support in Transformers. Full evaluation results, architecture details, training data, and usage examples are on each model card. IBM recommends pairing them with Granite Guardian for production deployments that need additional risk detection.