Mistral releases a new way of opening speakers

The French AI company Mistral released on Thursday a new type of speech recognition that can be used by AI voice assistants or businesses that use it as customer support. The model, which allows businesses to create voice assistants for sales and customer engagement, puts Mistral in direct competition with the likes of ElevenLabs, Deepgram, and OpenAI.

The new version, called Voxtral TTS, supports nine languages, including English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi, and Arabic.

“Our customers have been asking for a speaker model. So we created a small speaker model that can be connected to a smartwatch, a smartphone, a laptop, or other peripheral devices. Its price is a fraction of anything else on the market, but it offers high performance,” Pierre Stock, vp of scientific services at Mistral AI, told TechCrunch in a phone interview.

Mistral said the new model can edit a person’s voice with samples of less than five seconds, and capture features such as accent, intonation, intonation, intonation, and voice inaccuracies. Example, take Service 3Bthey can easily switch between languages ​​without losing the sound quality, which is useful for situations like translation or real-time translation. Stock said the company wanted the brand to sound human rather than robotic.

The model is designed for real-world applications, according to the company. It has a time-to-first-audio (TTFA) – a measure of the time it takes for the model to ‘speak’ after receiving input – of 90ms for a 10-second sample of 500 characters. The model also has a real-time response (RTF) of 6x, which means it can output a 10-second segment in about 1.6.

Earlier this year, Mistral launched two types of transcriptionone for large batch processing and the other for real-time applications with low latency. With the new communication system, the company aims to provide a complete range of voice services to businesses.

“We are planning to have the last platform that can use multimodal streams, including audio, text, and image and output as well. The main benefit is that you get many additional options with the end of the agetic system that supports audio as an input or output,” said Stock.

Mistral’s standout feature is that its open source and customization options enable businesses to adopt its voice over their competitors, as they can customize it to their liking.