Radar ·
Ant Ling open-sources five Ming-3.0 speech models for TTS, audio and ASR
Ant Ling announced the open-source release of five Ming-3.0 speech models, including TTS variants, audio and dialogue understanding models, and an ASR model for dialects. The source provides no links.
According to the briefing dated October 19, 2026, Ant Ling announces the release of five Ming-3.0 models as open source. They include two speech synthesis (TTS) variants: one with 0.5 billion parameters, featuring voice design, and another with 7.9 billion parameters, aimed at long-form speech. The list also includes audio and dialogue understanding models and an automatic speech recognition (ASR) model for dialects.
The source states that open-weight speech models for TTS and ASR can be hosted locally in voice products, subject to license details and benchmarks that the briefing does not describe. There are no links, performance figures or availability dates beyond the announcement. To verify, look for Ant Ling's original announcement and read the license and benchmark results they publish. This record does not reproduce results or claim performance. This news does not describe any Rota Nacional capability.
Original source: Ant Ling abre o código de cinco modelos de fala Ming-3.0 para TTS, áudio e ASR