Microsoft VibeVoice
Ranked 9 of 18 text-to-speech models by recommenders
Free, MIT-licensed Microsoft speech model praised for expressive output and built-in diarization; critics report random music artifacts and thin documentation.
Free, MIT-licensed Microsoft speech model praised for expressive output and built-in diarization; critics report random music artifacts and thin documentation.
26 people gave an opinion
- recommend it
- 10
- criticize it
- 16
- approval share
- 38%
Recommended in 4 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: recommended
- May 2026: not recommended
- June 2026: not recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: not recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 10 people recommend itNeeds 10 or more, met.
- 0.6 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 4 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Expressive, good-sounding speech
“It has some perks, is a bit more expressive in some cases, but overall is trained on really noisy data”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “In my experience, VibeVoice is quite good as well. Even the smaller model.”
Comment from 10 Nov 2025 in “TTS still sucks” “it's really great frontier open-source voice AI models”
Comment from 29 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” - Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI”
- Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI”
Built-in diarization and long transcription
“the built-in diarization is the one thing that actually caught my attention here”
Comment from 29 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “The 60-minute single-pass transcription is the part that actually matters.”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” Open-source with MIT license
“it's really great frontier open-source voice AI models”
Comment from 29 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “At least it's MIT licensed!”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI”
What people warn about
Each mark is one comment
Noisy training data, random music
“it seems this was trained on large amount of podcasts with ad jingles or phone call queues with elevator music”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “It has some perks, is a bit more expressive in some cases, but overall is trained on really noisy data”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “First of all 1.5B model documentation is not there.”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” Inconsistent or poor output quality
“the quantized version often goes off the rails and loses consistency in voice tone or accent”
Comment from 10 Nov 2025 in “TTS still sucks” “a slow, unoriginal product that doesn't actually do what it advertises”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “giving me very poor results”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” Missing documentation and frontend
“It would have been better if they provided not just weights, but also some frontend where it is usable as is.”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “First of all 1.5B model documentation is not there.”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” Openness claims; 7B model pulled
“MS is misrepresenting their openness here”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI” “they rug-pulled the 7B for our own safety”
Comment from 28 Apr 2026 in “Microsoft VibeVoice: Open-Source Frontier Voice AI”
Other text-to-speech models people recommend
The four most recommended other text-to-speech models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| ElevenLabs | 59 | 20 | 75% | 13 | — | Approved |
| Kokoro TTS | 59 | 9 | 87% | 11 | 1 mo3 owners | Approved |
| Pocket TTS | 41 | 16 | 72% | 8 | — | Approved |
| Qwen3-TTS | 41 | 13 | 76% | 7 | — | Approved |
| Microsoft VibeVoiceThis page | 10 | 16 | 38% | 4 | — | — |
What people said
26 of the 28 comments that mention it, newest first. Each links to the original comment.
“it's really great frontier open-source voice AI models”
“the built-in diarization is the one thing that actually caught my attention here”
“the SOTA is currently much more advanced”
“it seems this was trained on large amount of podcasts with ad jingles or phone call queues with elevator music”
“The 60-minute single-pass transcription is the part that actually matters.”
“still the most impressive local TTS model i've tried”
“Microslop really has a penchant for the worst names.”
“Saved a lot of my time thanks!”
“It has some perks, is a bit more expressive in some cases, but overall is trained on really noisy data”
“a slow, unoriginal product that doesn't actually do what it advertises”
“This is a very good model”
“you saved us a lot of time here.... i unstarred the repo”
“they rug-pulled the 7B for our own safety”
“MS is misrepresenting their openness here”
“It would have been better if they provided not just weights, but also some frontend where it is usable as is.”
“this is all built on stolen work”
“You just saved me an afternoon.”
“0.5B realtime is shit model. I was converting text, line by line and it was randomly adding music”
“First of all 1.5B model documentation is not there.”
“Microsoft is famous for choosing terrible names but how could they be this terrible.”
“At least it's MIT licensed!”
“giving me very poor results”
“Vibe Voice is the only real OSS cloning option.”
“Built entirely in Rust for performance and safety”
“In my experience, VibeVoice is quite good as well. Even the smaller model.”
“the quantized version often goes off the rails and loses consistency in voice tone or accent”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.