Speech to text is the automated transcription of spoken audio into written text using machine learning models. Truncated covers research developments, transcription accuracy benchmarks, software library releases, and hardware acceleration.
AI-generated summary Egyptian and Nigerian English have joined the Open ASR Leaderboard through a new collaboration with Liva AI. Liva AI says it collects multimodal interaction data through consumer products used by real people.
AI-generated summary AnyToSpeech is a web-based artificial intelligence text-to-speech and voice cloning platform available through PCWorld. A lifetime subscription to AnyToSpeech is currently on sale for $34.97 using the promo code SAVE5.
AI-generated summary Bayworks has launched an application that enables car dealership technicians to speak repairs aloud and automatically generate warranty claim reports. The software stores records of every repair and prompts technicians for missing details like measurements prior to submission.
AI-generated summary Android Authority has previewed the latest beta of the Gboard app, which reveals that the Gemini-powered voice typing feature Rambler could soon learn corrected words and contact names. The Gboard beta includes a new settings page and a Voice correction list where users can manage, import, export, or erase their saved vocabulary.
AI-generated summary Apple is currently testing new Audio Intelligence features in beta versions of iOS 27.3 and watchOS 27.3 for Apple Watch Series 12. The developer beta brings features like Siri Recaps to record and summarize conversations, raising both practical utility and privacy questions.
AI-generated summary Pebble launches the Qore Vox in India as a screenless wellness band that integrates a wrist microphone and an on-device AI transcription engine. The device provides continuous health monitoring and local audio transcription supporting English, seven Indian languages, and 24 foreign languages without requiring a subscription fee.
AI-generated summary Apple is beta testing its new Audio Intelligence technology for the Apple Watch, including conversation rewinding and recapping features for developers. Apple says the features will help users review conversations and take notes while maintaining privacy and omitting sensitive information.
AI-generated summary Nvidia has made available Parakeet, an open-source automatic speech recognition model that turns spoken audio into text locally without requiring cloud services. The model runs on modest hardware and can process recordings, voice notes, and web-based audio while keeping all data fully private on-device.
AI-generated summary Cartesia launched Cartesia Managed Agents as an end-to-end voice agent platform designed for enterprise conversations. Cartesia says the platform combines voice models and a harness to deliver natural speech-to-speech accuracy for enterprise tasks.
AI-generated summary NVIDIA releases Nemotron 3.5 Transcribe, an open-weight, promptable speech model now available on Hugging Face and ModelScope. NVIDIA says the model turns recorded speech into useful text by combining transcription, speaker attribution, translation, and audio question answering.
AI-generated summary Granola launched a dictation feature today that works on Mac just like any other such tool. Users can press the Fn key to start speaking while the tool cleans up and formats their speech across all apps.
AI-generated summary Baseten makes available an optimized implementation of Qwen3-ASR 1.7B for batch transcription. Baseten says the optimized stack transcribes 7x more audio per second at high concurrency compared with out-of-the-box vLLM.
AI-generated summary Cartesia makes its speech models available to Raise Financial Services for developing multilingual voice experiences within the Indian financial markets. Raise states that the partnership supports the shared ambition of making financial information more accessible through natural conversations.
The top AI news in your email delivered each morning. Choose categories