|

Alibaba’s new Qwen model to supercharge AI transcription tools

ASR error rates test of Alibaba Qwen

AI speech transcription tools are about to get much more aggressive with Alibaba’s Qwen crew pulling unveiling the Qwen3-ASR-Flash model.

Built upon the highly effective Qwen3-Omni intelligence and skilled utilizing an enormous dataset with tens of hundreds of thousands of hours of speech information, this isn’t simply one other AI speech recognition model. The crew says it’s designed to ship extremely correct efficiency, even when confronted with tough acoustic environments or complicated language patterns.

So, how does it stack up in opposition to the competitors? The efficiency information, from assessments carried out in August 2025, suggests it’s reasonably spectacular.

On a public check for normal Chinese, Qwen3-ASR-Flash achieved an error fee of simply 3.97 p.c, leaving opponents like Gemini-2.5-Pro (8.98%) and GPT4o-Transcribe (15.72%) trailing in its wake and displaying promise for extra aggressive AI speech transcription tools.

Qwen3-ASR-Flash additionally proved adept at dealing with Chinese accents, with an error fee of three.48 p.c. In English, it scored a aggressive 3.81 p.c, once more comfortably beating Gemini’s 7.63 p.c and GPT4o’s 8.45 p.c.

But the place it actually turns heads is in a notoriously tough space: transcribing music. 

When tasked with recognising lyrics from songs, Qwen3-ASR-Flash posted an error fee of simply 4.51 p.c, which is much better than its rivals. This skill to perceive music was confirmed in inner assessments on full songs, the place it scored a 9.96 p.c error fee; an enormous enchancment over the 32.79 p.c from Gemini-2.5-Pro and 58.59 p.c from GPT4o-Transcribe.

ASR error rates test of Alibaba Qwen's Qwen3-ASR-Flash comparing other popular AI speech recognition models used for transcription tools.

Beyond its spectacular accuracy, the model brings some modern options to the desk for next-generation AI transcription tools. One of the most important game-changers is its versatile contextual biasing.

Forget the times of painstakingly formatting key phrase lists, this method lets customers feed the model background textual content in nearly any format to get customised outcomes. You can present a easy listing of key phrases, whole paperwork, or perhaps a messy mixture of each. 

This course of eliminates any want for complicated preprocessing of contextual info. The model is sensible sufficient to use the context to sharpen its accuracy; but its basic efficiency is hardly affected even when the textual content you present is totally irrelevant.

It’s clear Alibaba’s ambition for this AI model is to grow to be a world speech transcription software. The service delivers correct transcription from a single model protecting 11 languages, full with quite a few dialects and accents.

The assist for Chinese is particularly deep, protecting Mandarin as well as to main dialects like Cantonese, Sichuanese, Minnan (Hokkien), and Wu.

For English audio system, it handles British, American, and different regional accents. The spectacular roster of different supported languages consists of French, German, Spanish, Italian, Portuguese, Russian, Japanese, Korean, and Arabic.

To spherical all of it out, the model can exactly determine which of the 11 languages is being spoken and is adept at rejecting non-speech segments like silence or background noise, making certain cleaner output than previous AI speech transcription tools.

See additionally: Siddhartha Choudhury, Booking.com: Fighting online fraud with AI

Banner for the AI & Big Data Expo event series.

Want to study extra about AI and massive information from trade leaders? Check out AI & Big Data Expo happening in Amsterdam, California, and London. The complete occasion is a part of TechEx and is co-located with different main know-how occasions, click on here for extra info.

AI News is powered by TechForge Media. Explore different upcoming enterprise know-how occasions and webinars here.

The submit Alibaba’s new Qwen model to supercharge AI transcription tools appeared first on AI News.

Similar Posts