Ant Group's Bailing open-sources five speech models in the Ming-3.0 series
Today, we're open-sourcing 5 Ming-3.0 models across 3 speech families.
AI BriefingToday, inclusionAI open-sourced five Ming-3.0 models across three speech families. Ming-3.0-tts includes a 0.5B speech design and paralinguistic-cue model, as well as a Tiny long-form speech model with 7.9B total parameters and 1.3B active parameters. Ming-3.0-audio includes a Flash understanding and tool-calling model with 126B total parameters and 6.5B active parameters, and a Tiny efficient dialogue model with 9.3B total parameters and 2.7B active parameters. Ming-3.0-asr-tiny is a dialect and context recognition model with 8.6B total parameters and 2B active parameters. The image adds that TTS supports seven languages plus two dialects, while ASR supports hotword lists and conversational context.
Also reported by 3 other sources
- Ant Group's Bailing open-sources the Ming-3.0 speech model series, covering speech synthesis, audio understanding and dialogue, and speech recognitionIT Home · Artificial Intelligence · 28 days ago
- Ant Group's Bailing open-sources the Ming-3.0 speech series, offering five models for speech synthesis, audio understanding, and speech recognitionWeChat Official Account: Ant Ling · 28 days ago
- Ant's inclusionAI open-sources the Ming-3.0-tts speech synthesis modelAnt Ling inclusionAI: New GitHub Repository · 28 days ago
Source: Ant Bailing · x.com
Event follow-up1
View all event coverage- Ant's inclusionAI open-sources the Ming-3.0-tts speech synthesis modelAnt Ling inclusionAI: New GitHub Repository · 28 days ago