Startup drops free multilingual speech model, taking aim at English-first AI bias.
Original headline: Reson8 releases Rumble-2-Flash speech model as a free open-weights download — siliconangle.com
More in AI and machine learning news
Amsterdam-based startup Reson8 has released Rumble-2-Flash, a 500-million-parameter speech recognition model, as a free open-weights download. The model is designed for fast transcription of prerecorded audio and supports over 40 languages, explicitly targeting the industry's English-centric bias. The company argues that dominant models are trained primarily on English, leaving millions of non-English speakers underserved, especially in specialized domains like medicine.
For developers and researchers, this release provides a practical, downloadable tool for building multilingual applications like meeting notetakers, voice agents, and captioning services. Benchmarks show it outperforms larger models like OpenAI's Whisper on lower-resource languages, while being optimized for real-world noisy audio. This move highlights a growing push for democratizing speech AI beyond just the world's most common language.
Read the original at siliconangle.com
Satirical commentary on real tech news, summarised with AI. Read the original source for the facts. How ZeroDaily works
