Gnani.ai has launched Vachana STT, a speech-to-text model built for Indian languages, under the IndiaAI Mission. The startup ...

Obsessing over model version matters less than workflow.

Brain activity during speech follows a layered timing pattern that matches large language model steps, showing how meaning builds gradually.

AI voice startup ElevenLabs today launched its Scribe v2 and Scribe v2 Realtime speech-to-text models designed for live, interactive applications. Scribe v2 delivers the highest possible accuracy in ...

Summary: New research shows that when people listen to speech at different speeds, the auditory cortex does not adjust its timing but instead processes sound in a fixed time window. This discovery ...

OpenAI launched the Realtime API in beta in October 2024. The API, which uses the same technology as ChatGPT’s advanced voice mode, enables software developers to create voice-based AI assistants that ...

New research shows models can be directly edited to hide selected voices, even when users specifically ask for them. A technique known as “machine unlearning” could teach AI models to forget specific ...

GenAI startup Sarvam AI, recently selected by the Centre to build India’s first homegrown LLM, has unveiled a new speech AI model that supports 11 Indian languages, including Punjabi, Marathi, Odia, ...

Nvidia has entered the open-source speech recognition arena with Parakeet-TDT-0.6B-v2, an automatic speech recognition (ASR) model now hosted on Hugging Face. Beyond its accuracy ranking, Nvidia ...

Abstract: Transformer and its derivatives have achieved success in diverse tasks across computer vision, natural language processing, and speech processing. To reduce the complexity of computations ...