AssemblyAI has introduced Universal-1, a powerful speech recognition model trained on over 12.5 million hours of multilingual audio data. It achieves best-in-class accuracy in English, Spanish, French, and German, outperforming competitors like OpenAI's Whisper-3 and NVIDIA Canary.
Assembly AI claims its new Universal-1 model has 30% fewer hallucinations than Whisper https://t.co/jfOTMmArgJ Visit https://t.co/l8fNQzV9nN for more AI news. #AI #artificialintelligence #audio #
Advancing ASR: AssemblyAI's Universal-1 Sets New Standards in Speech Recognition #accuracy #AI #AItechnology #ambientnoise #applications #artificialintelligence #ASR #AssemblyAI #Conformer #Efficiency #hallucinationrates #languages #llm https://t.co/9kDsha8LQ8 https://t.co/2Ruu4mbJyj
Breaking! 🚨 AssemblyAI launched Universal-1 ASR, a highly accurate, multilingual speech-to-text model. Key features: • Trained on 12.5 million hours (~3 petabytes) of training data • 30% less likely to make errors compared to Whisper • Costs $0.37 to process an hour of…
Automatic Speech Recognition is getting crazy good! Universal-1 is a new model pre-trained on 12.5 million hours of multilingual audio data and fine-tuned on 1.62 million human and pseudo-labeled data. And it's good! • It's 13.5% more accurate than other popular models •… https://t.co/EiUzl8OWz3
AssemblyAI Unveils Universal-1: Surpassing Whisper-3 with Groundbreaking Accuracy and Speed in Speech Recognition Quick read: https://t.co/WUWL2vK3oE Try Universal-1 on Playground: https://t.co/76kXmnjQ3J Key Takeaways: ✅ Universal-1 outperforms OpenAI’s Whisper-3, offering…
Today @AssemblyAI announced its most powerful speech recognition model to date: Universal-1. Trained on over 12.5 million hours of multilingual audio data, it achieves best-in-class speech-to-text accuracy. #AccelFamily https://t.co/on2y91KP7x
Biggest update to AI Speech Recognition Model Assembly AI just released an highly accurate, multilingual speech-to-text AI model Universal-1 It achieves the best accuracy across English, Spanish, French, German & beats OpenAI whisper, NVIDIA Canary on a number of benchmarks. https://t.co/ZWomv6AraJ
Wow. While OpenAI API is still stuck on Whisper-2, @AssemblyAI releases something that beats even Wishper-3: + 13.5% more accurate than Whisper-3 + Up to 30% fewer hallucinations + 38s to process 60min audio - Only 20 languages supported Free playground https://t.co/1hNGinI1oW https://t.co/6NJMsEcGGt https://t.co/7eS3Dy5BbU
Introducing Universal-1, our most powerful speech recognition model to date. Trained on over 12.5 million hours of multilingual audio data, Universal-1 achieves best-in-class speech-to-text accuracy across English, Spanish, French, and German. https://t.co/KYdlu9a43W https://t.co/xCGPezm1L0