It has some perks, is a bit more expressive in some cases, but overall is trained on really noisy data, uses more memory, and isn't that fast - I'm talking about the (7b?) version that they released then removed quickly (vibevoice-community on github) - I still use chatterbox turbo and sometimes qwen TTS.
there is so much more subversive marketing out there than any of us can really fathom. i try not to be too paranoid but it's getting a lot harder every day.
i know someone who worked in what we might call the 'astroturfing' space within the entertainment industry. after having a few discussions with him and with things like this[0] becoming more known, it's really difficult to afford any assumption of organic intent when money is on the line - especially at the scale that microsoft works at compared to something as comparatively quaint as the music industry.
It is not good for text to speech (TTS) as well. I am trying it for few days. First of all 1.5B model documentation is not there. 0.5B realtime is shit model. I was converting text, line by line and it was randomly adding music and couldn't handle special characters like "…".
I really disappointed with this model to say the least.
The 7B parameter Vibevoice TTS model is still the most impressive local TTS model i've tried. It was pulled by Microsoft a few days after its release due to "abuse potential" but it can be found in various community maintained huggingface repos.
yep, it seems this was trained on large amount of podcasts with ad jingles or phone call queues with elevator music. I was also pretty disappointed to run the TTS last week.
Devils advocate here: I can give you a binary of my open source MIT code and never phone you the code. The code is still MIT licensed, and open source. You just have no access to it.
That said, I entirely agree that MS is misrepresenting their openness here, which isn’t in the least surprising.
In their defense, most everyone else does the same thing. They still shouldn't do it, but at least they're not the trendsetter here (though they are contributing to the ongoing problem)
I mean, you have "AI" which means just about anything in marketing speak, "Agentic" is kind of becoming similar, hopefully they don't goof that one too badly, would be nice to know what you are trying to sell me. Used to be "Cloud" meant storage not just hosting (I guess it still does).
Then there's "Smart" in front of Car, Phone, TV, and so on... Meaning different things.
I do think "Open Weight" should be more commonly used. There's definitely communities that spring up that build the training infrastructure and inference infrastructure around open models on the other hand.
Open weights is not exactly right either because we do get source of the software that uses those open weights.
Maybe open inference?
But we often also get source code for fine tunning the model.
So maybe it's closer to open source than to anything else?
Isn't it a bit like not calling a game open source because engine tooling used to made it isn't open source and they didn't publish .psd files with asset designs?
I'm genuinely torn on this one; I get technically why not, but why I think I have no problem with it is the wishy-washiness of "open source" generally.
As I teach this stuff to people newer to this tech, it's probably just easier and more helpful to refer to the wide array of "stuff you can just download and use yourself" as "open-source" and then after that, go deeper and talk about why Stallman was right, how "Free Software" was first. etc.
Look at the "News" section in the readme - The original TTS model is gone from this repo (you can still find it other places), but the SST/ASR, long form TTS, and streaming TTS models are newer.
When explanations get posted directly in HN comments, I imagine someone somewhere in the world is able to learn in spite of their Internet restrictions/firewalls
People will also post their own interpretations in response to comments, and quickly find out they missed something.
… But if you try to automate it, like include a summary under every HN post, you encourage laziness too much and are pre-chewing too heavily. Some balance here.
[on topic]
(OK I’m done making excuses, time to read the article… thanks for the encouragement!)
I thought this was not explained in the readme directly but in fact I missed it. I wasn’t going to read Microsoft entire changelog! But it was substantive, thanks to sibling commenter:
“2025-09-05: VibeVoice is an open-source research framework intended to advance collaboration in the speech synthesis community. After release, we discovered instances where the tool was used in ways inconsistent with the stated intent. Since responsible use of AI is one of Microsoft’s guiding principles, we have removed the VibeVoice-TTS code from this repository.”
In my mind, Vibe-anything means "some slop carelessly thrown together to ship as fast as possible." Wild that it's being used in a serious product name!
"get offended" is just what the clickbait news cycle made of it. It was based on the post at [1], and this is all it said:
> We need to get beyond the arguments of slop vs sophistication and develop a new equilibrium in terms of our “theory of the mind” that accounts for humans being equipped with these new cognitive amplifier tools as we relate to each other
I built speech-swift, which focuses on on-device speech processing like VibeVoice, but specifically leverages Apple Silicon's capabilities for ASR, TTS, and VAD without cloud dependency. Our ASR supports 52 languages with a real-time factor of 0.06. https://soniqo.audio/benchmarks
Note that this just covers the Speech-to-Text/Speech-Recognition aspect (a-la whisper), there's also models for long-form Text-To-Speech and steaming Text-To-Speech.
I'd be willing to bet it will be "Word of the Year" for 2026. Merriam-Webster had 'slop' for 2025, and 'polarization' for 2024. Is there a prediction market for this?
I've been using VibeVoice's ASR (speech to text) model quite intensively for the past month and have found it to be a lot more reliable and out-of-the box functional then Whisper, parakeet and other models. The fact that is has diarization built into to the model is a huge win in my book. Without that you have to run a different model just for that which adds significantly to the overall processing time vs VibeVoice which gives you reliably great results. Big fan.
The 60-minute single-pass transcription is the part that actually matters. Most ASR models chunk audio and you lose speaker continuity across boundaries. If the diarization actually holds up on hour-long recordings without drifting, thats a real solve for podcast and meeting transcription workflows.
I took a look into local options for ASR and diarization some months ago, I missed that VibeVoice now has this feature.
My conclusions back then (which only came from a shallow research on the topic and 0 real experience mind you) was that Whisper + Pyannote was the "stable" approach.
Have the VibeVoice, Voxtral, Qwen or the Nemo solutions caught up in segmentation and speaker recognition?
It highly depends on the sort of data you’re processing (phone calls, podcasts, meetings of more people recorded using single channel?). For NVIDIA/NeMo, check out their softformer diarization models (also streaming).
the built-in diarization is the one thing that actually caught my attention here. running whisper + pyannote separately is a pain for long recordings and the speaker continuity breaks at chunk boundaries. if this handles it in a single pass that's a real workflow improvement, regardless of how the raw accuracy benchmarks compare
Local? No idea. Cloud? Eleven Labs, probably. But it's described as "cloning" not "training". Not sure what the distinction is or why it matters if the end result is you can to generate any TTS that sounds like you. There might very well be an important one, I just don't know it.
Seems quite heavy for a STT model, Parakeet and Whisper are much smaller and perform great for quick dictation and transcription of longer files. I guess that's due to additional accuracy and speaker diarisation?
The TTS example clip in the repo of 'spontaneous singing' is creepy as fuck
This is not a new model. Also, it hallucinates a lot. Also, it's very heavy and slow in inference. It's also bad in multilingual.
Edit: I'm talking purely about speech to text (STT). Not sure about the other things this can do.
It has some perks, is a bit more expressive in some cases, but overall is trained on really noisy data, uses more memory, and isn't that fast - I'm talking about the (7b?) version that they released then removed quickly (vibevoice-community on github) - I still use chatterbox turbo and sometimes qwen TTS.
Yeah, I don't get why it is suddenly getting so much attention today, it is all over twitter too
Simonw (who has a bit of a Midas touch for posts here) just posted about it https://simonwillison.net/2026/Apr/27/vibevoice/
3 replies →
there is so much more subversive marketing out there than any of us can really fathom. i try not to be too paranoid but it's getting a lot harder every day.
i know someone who worked in what we might call the 'astroturfing' space within the entertainment industry. after having a few discussions with him and with things like this[0] becoming more known, it's really difficult to afford any assumption of organic intent when money is on the line - especially at the scale that microsoft works at compared to something as comparatively quaint as the music industry.
[0] https://www.wired.com/story/geese-chaotic-good-marketing-ind...
well duh, they updated the news section
https://github.com/microsoft/VibeVoice/commit/e73d1e17c3754f...
which is microsoft for "we removed two dead links". AI innovation knows no limits!
1 reply →
It is not good for text to speech (TTS) as well. I am trying it for few days. First of all 1.5B model documentation is not there. 0.5B realtime is shit model. I was converting text, line by line and it was randomly adding music and couldn't handle special characters like "…".
I really disappointed with this model to say the least.
The 7B parameter Vibevoice TTS model is still the most impressive local TTS model i've tried. It was pulled by Microsoft a few days after its release due to "abuse potential" but it can be found in various community maintained huggingface repos.
yep, it seems this was trained on large amount of podcasts with ad jingles or phone call queues with elevator music. I was also pretty disappointed to run the TTS last week.
Yes, the SOTA is currently much more advanced.
What do you consider to be SOTA?
you saved us a lot of time here.... i unstarred the repo
moving on....
I don't really pay attention to stars. Do people use them as bookmarks? Why would you star a repo if you knew so little about it?
3 replies →
Saved a lot of my time thanks!
I'm shocked, shocked to find that Microsoft takes credit for a slow, unoriginal product that doesn't actually do what it advertises.
Imagine the balls it took to willingly attach the Microsoft label to the front of the product that is Teams.
3 replies →
You just saved me an afternoon.
[flagged]
The nuance is lost on LLM agentic dominant partakers.
I think we should stop calling this type of models open source. They are indeed "open weight." The training code is proprietary and never revealed.
https://github.com/microsoft/VibeVoice/issues/102
Indeed. We now live in a world where freeware is named open source. We are very sorry, Stallman.
If you're going to apologize to Stallman, you should apologize for conflating open source with software freedom. ;D
10 replies →
I'm reserving that complaint for "open source" models which are released under non-open-source licenses.
I care that I know what I can DO with the project when I see it described as "open source".
> I care that I know what I can DO with the project when I see it described as "open source".
Yes, the first of which is that you should be able to build it from source. Which requires the source code, and in this case data.
6 replies →
That would be “permissive license”
Maybe we should have a little cue card for models: vendor/name, size, open weights, open source, permissive license.
It’s simple enough an idea.
> we should stop calling this type of model open source. They are indeed "open weight”
This ship has sailed. It’s now in the same category as hacker/cracker and the pronunciation of GIF.
I think you mean GIF.
The inventor of GIF didn't begin with a document* clearly laying out what is and isn't to be called a "GIF."
I think it's right to push back whenever a huge tech corporation tries to build goodwill by falsely using terms like "open source."
*https://opensource.org/osd
3 replies →
It's the same as GIS, you wouldn't say jizz now would you?
18 replies →
And "hallucination" which should have been "delusion".
Way early on (spring 2023) people tried to stop it, but no luck.
2 replies →
Devils advocate here: I can give you a binary of my open source MIT code and never phone you the code. The code is still MIT licensed, and open source. You just have no access to it.
That said, I entirely agree that MS is misrepresenting their openness here, which isn’t in the least surprising.
? Do you know what “source” means in open source? Like, what is the source of the binary? It’s the code. That’s the source in open source.
4 replies →
In their defense, most everyone else does the same thing. They still shouldn't do it, but at least they're not the trendsetter here (though they are contributing to the ongoing problem)
At least it's MIT licensed! As much as non-open training data irks me, restrictive licensing irks me more!
what is problem with restrictive licensing? Most of them starts if you have 1M users etc?
What you said makes a lot of sense. Free software should not be confused with open source
I mean, you have "AI" which means just about anything in marketing speak, "Agentic" is kind of becoming similar, hopefully they don't goof that one too badly, would be nice to know what you are trying to sell me. Used to be "Cloud" meant storage not just hosting (I guess it still does).
Then there's "Smart" in front of Car, Phone, TV, and so on... Meaning different things.
I do think "Open Weight" should be more commonly used. There's definitely communities that spring up that build the training infrastructure and inference infrastructure around open models on the other hand.
Open weights is not exactly right either because we do get source of the software that uses those open weights.
Maybe open inference?
But we often also get source code for fine tunning the model.
So maybe it's closer to open source than to anything else?
Isn't it a bit like not calling a game open source because engine tooling used to made it isn't open source and they didn't publish .psd files with asset designs?
I'm genuinely torn on this one; I get technically why not, but why I think I have no problem with it is the wishy-washiness of "open source" generally.
As I teach this stuff to people newer to this tech, it's probably just easier and more helpful to refer to the wide array of "stuff you can just download and use yourself" as "open-source" and then after that, go deeper and talk about why Stallman was right, how "Free Software" was first. etc.
[dead]
Openwashing is the new greenwashing, which, coincidently, seems to have gone out of fashion a few hundred datacentres ago.
it was replaced with abundancewashing
2 replies →
I think in this category, Voxtral by Mistral is a lot better. It also happens to be small enough to run on webGPU https://huggingface.co/spaces/mistralai/Voxtral-Realtime-Web...
Interesting story about this repo/product/author by cybersecurity researcher Kevin Beaumont: https://cyberplace.social/@GossiTheDog/116454846703138243
got to love how they're trying to hide the links.
Isn't this project the one Microsoft published but then soon after pulled it for security/safety reasons? What has changed since then?
Look at the "News" section in the readme - The original TTS model is gone from this repo (you can still find it other places), but the SST/ASR, long form TTS, and streaming TTS models are newer.
[flagged]
It’s confusing (at least for me) because the project covers a number of things including what you are mentioning.
[off topic]
When explanations get posted directly in HN comments, I imagine someone somewhere in the world is able to learn in spite of their Internet restrictions/firewalls
People will also post their own interpretations in response to comments, and quickly find out they missed something.
… But if you try to automate it, like include a summary under every HN post, you encourage laziness too much and are pre-chewing too heavily. Some balance here.
[on topic]
(OK I’m done making excuses, time to read the article… thanks for the encouragement!)
I thought this was not explained in the readme directly but in fact I missed it. I wasn’t going to read Microsoft entire changelog! But it was substantive, thanks to sibling commenter:
“2025-09-05: VibeVoice is an open-source research framework intended to advance collaboration in the speech synthesis community. After release, we discovered instances where the tool was used in ways inconsistent with the stated intent. Since responsible use of AI is one of Microsoft’s guiding principles, we have removed the VibeVoice-TTS code from this repository.”
Interesting to see "vibe" enshrined by the likes of Microsoft as an AI product word.
Especially when "vibe coded" can have a negative connotation meaning quickly put together without understanding.
In my mind, Vibe-anything means "some slop carelessly thrown together to ship as fast as possible." Wild that it's being used in a serious product name!
I’m just surprised they put the name of the e-waste slop company in their product
Maybe they were trying to make a pun on "Via Voice", the cursed IBM STT from the 90s?
I'm honestly more surprised that they could resist the temptation to call it Copilot
Microslop Copilot for Voice! After they renamed Office, they surely will rename this one, too.
[flagged]
"get offended" is just what the clickbait news cycle made of it. It was based on the post at [1], and this is all it said:
> We need to get beyond the arguments of slop vs sophistication and develop a new equilibrium in terms of our “theory of the mind” that accounts for humans being equipped with these new cognitive amplifier tools as we relate to each other
[1] https://snscratchpad.com/posts/looking-ahead-2026/
2 replies →
I built speech-swift, which focuses on on-device speech processing like VibeVoice, but specifically leverages Apple Silicon's capabilities for ASR, TTS, and VAD without cloud dependency. Our ASR supports 52 languages with a real-time factor of 0.06. https://soniqo.audio/benchmarks
Great post last night from Simon: https://simonwillison.net/2026/Apr/27/vibevoice/
Note that this just covers the Speech-to-Text/Speech-Recognition aspect (a-la whisper), there's also models for long-form Text-To-Speech and steaming Text-To-Speech.
“VibeVoice can only handle up to an hour of audio”
Why?
You have selected Microsoft Sam as the computer's default voice.
My friends and I had fun in the computer lab with Microsoft Sam, inputting long strings of characters to create funny sound effects. Sususususususu.
Holy moly, a Microsoft AI product that isn't named Copilot!
Missed opportunity to call it Vopilot
Slopilot
So we've really just settled on Vibe as the verb for AI then?
I'd be willing to bet it will be "Word of the Year" for 2026. Merriam-Webster had 'slop' for 2025, and 'polarization' for 2024. Is there a prediction market for this?
it'll probably be something we're not even talking about yet - we still have 7 months in which to make the world even worse
Why use precise technical language when you can just vibe with your AI system?
Still waiting for the open weights model that conclusively beats the multi-year old Whisper in accuracy, features, and performance.
It's crazy that a lot is happening in open models for stt, but there's very little progress when it comes to results, esp multilingual.
Surprised it wasn't called Copilot Voice
Microsoft Store App Vibing.exe Accused of Harvesting Screens, Audio, and Clipboard Data:
https://cyberpress.org/microsoft-store-app-vibing-exe-accuse...
I've been using VibeVoice's ASR (speech to text) model quite intensively for the past month and have found it to be a lot more reliable and out-of-the box functional then Whisper, parakeet and other models. The fact that is has diarization built into to the model is a huge win in my book. Without that you have to run a different model just for that which adds significantly to the overall processing time vs VibeVoice which gives you reliably great results. Big fan.
The 60-minute single-pass transcription is the part that actually matters. Most ASR models chunk audio and you lose speaker continuity across boundaries. If the diarization actually holds up on hour-long recordings without drifting, thats a real solve for podcast and meeting transcription workflows.
I took a look into local options for ASR and diarization some months ago, I missed that VibeVoice now has this feature.
My conclusions back then (which only came from a shallow research on the topic and 0 real experience mind you) was that Whisper + Pyannote was the "stable" approach.
Have the VibeVoice, Voxtral, Qwen or the Nemo solutions caught up in segmentation and speaker recognition?
It highly depends on the sort of data you’re processing (phone calls, podcasts, meetings of more people recorded using single channel?). For NVIDIA/NeMo, check out their softformer diarization models (also streaming).
the built-in diarization is the one thing that actually caught my attention here. running whisper + pyannote separately is a pain for long recordings and the speaker continuity breaks at chunk boundaries. if this handles it in a single pass that's a real workflow improvement, regardless of how the raw accuracy benchmarks compare
I the past month or so, I added 2 models to my app Whisper Memos (https://whispermemos.com):
- Cohere Transcribe (self hosted)
- Grok Speech To Text (they provide an API, only $0.10/hr!)
They are both excellent. I'm not sure about this one. Would you like to see it in a consumer speech to text app?
I've had good experiences with the Mistral Voxtral models (I've used the API, but some of the model-variants are open weight)
Does Cohere work with longer transcripts? Do you have to do some magic to merge recordings over 35 seconds long?
Have you tried qwen?
Any non-Musk alternatives that are comparable in quality and cost?
Voxtral competes on price ($0.003/min) and quality. Speechmatics has best in class accuracy but is a bit more expensive ($0.004/min)
Our default is still OpenAI Whisper. Grok is just a choice for users who might prefer it.
What’s the current state of the art, for each of training locally and in the cloud, for learning my voice?
Locally maybe https://voicebox.sh/
Elevenlabs in the cloud.
Local? No idea. Cloud? Eleven Labs, probably. But it's described as "cloning" not "training". Not sure what the distinction is or why it matters if the end result is you can to generate any TTS that sounds like you. There might very well be an important one, I just don't know it.
open weights i would say S2: https://github.com/rodrigomatta/s2.cpp
Microsoft has historically made poor choices in product naming, but this has to be a new low.
Shouldn't it be called something like "Copilot Voice"?
That's not confusing enough. It should be just Copilot.
Microsoft continues to be completely incapable of coming up with good names for their products and services
Someone tell me if this is better or worse than Parakeet
When mixing languages, why does the English have Chinese accent and Chinese have English accent? Is it a feature or bug?
Maybe Microsoft’s real strength was never making the best model, it was knowing you don’t need to, as long as you own the platform everyone builds on.
https://www.youtube.com/watch?v=d_AP3SGMxxM
Explains most of the shit they have pushing with Windows 11. Perhaps all that bloatware was VibeVoiced too.
It would have been better if they provided not just weights, but also some frontend where it is usable as is.
Seriously, VibeVoice? Microslop really has a penchant for the worst names.
For me its giving me very poor results
looks like this offers ASR support in GGUF https://github.com/CrispStrobe/CrispASR -- haven't tested
This is a very good model, but can it be run on the web?
What the do they mean by frontier voice
Isn't voxtral much better?
Sounds like Msft wanted to coast on the “vibecode” vibe popularity?
Seems quite heavy for a STT model, Parakeet and Whisper are much smaller and perform great for quick dictation and transcription of longer files. I guess that's due to additional accuracy and speaker diarisation?
The TTS example clip in the repo of 'spontaneous singing' is creepy as fuck
English only?
Previously:
Sept 2025 https://news.ycombinator.com/item?id=45114245
That was about the text-to-speech model, the speech-to-text one was release in January.
Microsoft is famous for choosing terrible names but how could they be this terrible.
What a terrible name
lol they rug-pulled the 7B for our own safety some months ago
[flagged]