AssemblyAI’s Post

Congratulations to the AssemblyAI team on the launch of their new speech AI model, Universal-1. This announcement underscores our shared commitment to continually raise the bar on transcription accuracy, ensuring our customers receive the most reliable and actionable insights from their conversations. Trained on 12.5M total unique hours of multilingual audio data, Universal-1 hallucinates up to 30% less and is 22% more accurate than other APIs. For CallRail Conversation Intelligence®️ customers, more accurate transcripts mean more accurate call summaries, sentiment analysis, self-reported attribution, keyword spotting, and call coaching opportunities. In short — more valuable insights and more confident marketing! Read more about AssemblyAI’s newest model. 👇🏼

AssemblyAI is one of the coolest companies developing speech products. Universal-1 is an incredible outcome (I genuinely hope it becomes open-source eventually), both from an engineering and research perspective. Kudos to the entire team for the hard work! However, at the same time, this shows us how speech recognition (and any speech-related task) and AI are not even close to being solved yet. Using 12.5M hours of data is huge! Almost 1400 years of continuously listening to utterances. If you are afraid of how fast AI development is and think there is little to explore these days, this is the best time and a great opportunity to jump in.

AssemblyAI released Universal 1, their most powerful and accurate multilingual speech-to-text model to date—trained on 12.5M hours of multilingual audio data. This model is swift and easy for developers to integrate into their workflow. More about it here: https://lnkd.in/dZPqjsr8 #ai #ml #texttospeech #developer #assemblyai

AssemblyAI released Universal 1, their most powerful and accurate multilingual speech-to-text model to date—trained on 12.5M hours of multilingual audio data. This model is swift and easy for developers to integrate into their workflow. More about it here: https://lnkd.in/dnAywR7g #ai #ml #texttospeech #developer #assemblyai

Some exciting updates to share about Unmixr AI: 1. Added support for locales for all multilingual voice in the voice studio 2. Onboarded two new multilingual voices 3. Enhanced Audio Merge Studio 4. Document Translation is in private preview, will be open to public soon! Unmixr AI

Language barriers in customer support present a widespread challenge for organizations globally. Spotify overcame this by leveraging Sutherland Translate AI, enabling seamless real-time multi-language support with native language nuances and increasing their service capacity by 30% across 4 languages! Find out how this AI-powered platform is powering consistent and culturally nuanced customer experiences for Spotify. Read Case Study: https://bit.ly/3WzRU4L #SutherlandDES #AI

Imagine conversing seamlessly with a local merchant in a bustling night market, surrounded by chatter in a foreign language. Long-distance speech recognition can help overcome language barriers with the incorporation of AI technology, even within noisy settings. Check out the article below for more and visit us at Yobe Inc. for a deeper insight into the future of voice technology. #YobeInc #VoiceAI #VoiceTechnology #VoiceTech #SpeechRecognition https://lnkd.in/eRPEgFzM

In this article - https://bit.ly/3v47j1H - 5 considerations are outlined when buying Speech AI. We have everything you need to overcome these potential challenges. 1. Language Barriers With consistently high accuracy across 49 languages and counting, Speechmatics removes language barriers like never before - https://bit.ly/471MVvA 2. Enhancing Accuracy Earlier this year, we launched Ursa, the world’s most accurate speech-to-text system. Ursa delivers unprecedented performance across a diverse range of voices - https://bit.ly/3uXq4UK 3. Breaking Through Background Noise We put all of our tech through rigorous testing to ensure it will work in even challenging environments. In a test taken from audio recorded in a moving car, Speechmatics made 44.48% fewer errors than Whisper - https://bit.ly/3sevFVM 4. Adapting to Industry Lingo Speechmatics built Custom Dictionary for precisely this reason. Boost accuracy for proper nouns, acronyms or industry-specific terms by providing a list of custom words - https://bit.ly/4afevZ3 5. Tailored Solutions No two customers are the same. Our sales and solutions engineers would love to talk to you about how to tailor Speechmatics to drive the most value for your business - https://bit.ly/3uUYkjr