Text-to-Speech

Evidence (6)

• 7Hugging Face Blog·9/30/2026research
  Hugging Face Launches Open TTS Leaderboard for Multilingual Evaluation

  Hugging Face introduced the Open TTS Leaderboard to evaluate open-source and multilingual text-to-speech models with objective metrics. It measures intelligibility, speed, and speaker similarity, aiming to scale evaluation beyond human-vote arenas while still complementing them.

• 2Simon Willison·9/23/2026research
  Gemini 3.8 adds text-to-speech greeting

  Google DeepMind says Gemini 3.8 can generate text-to-speech output that says hello. The announcement indicates a new speech capability in the Gemini 3.8 model, but provides no further technical details in the headline alone.

• 6Hugging Face Blog·8/10/2026model_release
  NVIDIA Magpie TTS adds open multilingual voice support

  NVIDIA Magpie Multilingual TTS is presented as an open-weights text-to-speech model for building low-latency voice agents with full deployment control. The latest release expands coverage to Modern Standard Arabic, Korean, and Brazilian Portuguese, while also improving quality and code-switching support across existing languages.

• 6Hacker News·8/5/2026research
  Former OpenAI researcher joins Conduit to build thought-to-text telepathy

  A former OpenAI researcher says they resigned and joined Conduit as a founding researcher to work on thought-to-text systems, which the company frames as non-invasive telepathy using neural data. The post outlines a vision for how people may interact with AI through headband-based brain interfaces over the next few years and says Conduit is collecting large amounts of non-invasive neural data to support the effort.

• 7Hugging Face Blog·7/15/2026research
  Hume AI unveils Real World VoiceEQ for evaluating voice AI quality

  Hume AI introduced Real World VoiceEQ, a benchmark for measuring the human quality of voice AI across speech recognition, text-to-speech, speech-to-speech, and speech understanding. Built from over 1 million human ratings, it evaluates more than 40 models across 15+ dimensions and suggests current voice systems are increasingly specialized rather than having a single best performer.

• 5Martin Fowler·6/16/2026research
  How Bayer Built a Reliable Agentic AI System for Preclinical Research

  This case study describes PRINCE, a cloud-hosted platform built by Bayer and Thoughtworks to improve access to preclinical safety study data. The system evolved from keyword search into an agentic retrieval-augmented generation and Text-to-SQL assistant that can answer complex questions and draft regulatory documents, with engineering focused on transparency, recovery, observability, and human oversight.