IBM Granite Speech 5.0 2B AI Model Released
IBM launches Granite Speech 5.0 2B, an Apache 2.0 speech model ranking #2 on Open ASR Leaderboard with 5.68% average English WER.
Stock photo for illustration only, not from the actual event
- IBM releases Granite Speech 5.0 2B for multilingual speech recognition and translation under Apache 2.0.
- Ranks #2 on the Open ASR Leaderboard with an average English Word Error Rate (WER) of 5.68%.
- Launched alongside updated Turbo CTC and 4.1 2B Plus models.
IBM has officially introduced its newest artificial intelligence model named Granite Speech 5.0 2B, designed specifically for multilingual transcription and translation tasks. The model is released under the Apache 2.0 license, allowing developers and organizations open access for integration and deployment.
The new model has demonstrated exceptional performance on the global stage, securing the #2 position on the Open ASR Leaderboard, which serves as a benchmark for Automatic Speech Recognition systems. It achieved an impressive average English Word Error Rate (WER) of just 5.68%.
Stock photo for illustration only, not from the actual event
The release of Granite Speech 5.0 2B under the Apache 2.0 license highlights IBM's continued commitment to open-source AI, empowering developers to leverage advanced speech processing technology. Achieving a low WER of 5.68% on the Open ASR Leaderboard indicates robust accuracy in handling complex vocabulary and diverse accents, which is crucial for applications such as customer service automation, meeting transcription, and real-time translation.
Alongside the flagship Granite Speech 5.0 2B model, IBM also rolled out updates and additions to its related model lineup, including:
- The Turbo CTC model, optimized for faster processing speeds.
- The 4.1 2B Plus model, expanding the capabilities of the audio processing suite.
This expansion of the Granite model family represents another significant milestone for IBM in strengthening its natural language processing and speech technology offerings for global enterprises and developers.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment