text-to-speech

Microsoft AI

2026-10-02 09:31:52

Microsoft AI rolls out two speech models for live transcription and voice agents

Microsoft AI has introduced two new models, MAI-Transcribe-2-Streaming and MAI-TTS-2, aimed at improving capabilities for voice agents. According to a Techub News item citing The Decoder, the first model focuses on real-time speech-to-text transcription, while the second converts text into natural-sounding speech. Both were tuned for voice-agent use cases rather than presented as general-purpose releases. The announcement centers on two core functions that sit at the heart of spoken AI systems: converting live audio into text with low latency, and generating spoken responses from written input. In the brief release, Microsoft AI framed the pair as tools designed to strengthen speech-based agent experiences. No additional technical specifications, launch regions, pricing details, or deployment timelines were disclosed in the source item.

70

Gradium AI

2026-09-01 03:42:34

Gradium AI Rolls Out New Default TTS Model With 81.0% Pass Rate on Hard Cases

Gradium AI has released a new text-to-speech model and made it the default for its API and Studio, according to Techub News. The announcement, first reported by MarkTechPost, reveals that the model achieved an 81.0% human-rated pass rate on a hard-case test set made up of 500 sentences across five languages. That result places it ahead of Cartesia Sonic 3.6, which posted 75.1%, and ElevenLabs v3 Conversational, which came in at 65.4%. In the Coval benchmark, the median time to first audio generation is 216 milliseconds, which the company says is 170 milliseconds faster than the alternative model it replaces. The new model has been live and set as default since August 31, 2026, meaning existing users do not need to migrate or make any changes. Gradium AI has also open-sourced the full 500-sentence evaluation dataset on Hugging Face, so that the test data is publicly available. The update covers both the API and the Studio platform, making the model the standard choice for new requests.

260