44 comments

  • toebee 0 minutes ago
    We have a ZeroGPU Space provided by HuggingFace up and running! Test it now on https://huggingface.co/spaces/nari-labs/Dia-1.6B
  • hemloc_io 5 hours ago
    Very cool!

    Insane how much low hanging fruit there is for Audio models right now. A team of two picking things up over a few months can build something that still competes with large players with tons of funding

    • toebee 1 hour ago
      Thank you for the kind words <3
  • tyrauber 6 hours ago
    Hey, do yourself a favor and listen to the fun example:

    > [S1] Oh fire! Oh my goodness! What's the procedure? What to we do people? The smoke could be coming through an air duct!

    Seriously impressive. Wish I could direct link the audio.

    Kudos to the Dia team.

    • jinay 4 hours ago
      For anyone who wants to listen, it's on this page: https://yummy-fir-7a4.notion.site/dia
      • mrandish 3 hours ago
        Wow. Thanks for posting the direct link to examples. Those sound incredibly good and would be impressive for a frontier lab. For two people over a few months, it's spectacular.
      • DoctorOW 3 hours ago
        A little overacted, it reminds me of the voice acting in those flash cartoons you'd see in the early days of YouTube. That's not to say it isn't good work, it still sounds remarkably human. Just silly humans :)
        • selimthegrim 1 hour ago
          Reminded me of the Fenslerfilm G.I. Joe sketch where the kids have something on the stove burning
          • wisemang 1 hour ago
            Stop all the downloading!
    • 3abiton 2 hours ago
      This is oddly reminiscent of the office. I wonder if tv shows were part of its training data!
    • toebee 1 hour ago
      Thank you!! Indeed the script was inspired from a scene in the Office.
    • nojs 3 hours ago
      This is so good. Reminds me of The Office. I love how bad the other examples are.
  • rustc 4 hours ago
    Is this Apache licensed or a custom one? The README contains this:

    > This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

    > This project offers a high-fidelity speech generation model *intended solely for research and educational use*. The following uses are strictly forbidden:

    > Identity Misuse: Do not produce audio resembling real individuals without permission.

    > ...

    Specifically the phrase "intended solely for research and educational use".

    • toebee 35 minutes ago
      Sorry for the confusion. the license is plain Apache 2.0, and we changed the wording to "intended for research and educational use." The point was, users are free to use it for their use cases, just don't do shady stuff with it.

      Thanks for the feedback :)

    • montroser 1 hour ago
      Hmm, the "strictly forbidden" part seems more important than whatever are their stated intentions... Either way, it seems like it needs clarifying.
  • notdian 6 hours ago
    made a small change and got it running on M2 Pro 16GB Macbook pro, the quality is amazing.

    https://github.com/nari-labs/dia/pull/4

    • toebee 7 minutes ago
      Thank you for the contribution! We'll be merging PRs and cleaning code up very soon :)
    • rahimnathwani 25 minutes ago
      Thank you for this! My desktop GPU has only 8GB VRAM, but my MacBook has plenty of unified RAM.
    • noiv 3 hours ago
      Can confirm, runs straight forward on 15.4.1@M4, THX.
  • Havoc 3 hours ago
    Sounds really good & human! Got a fair bit of unexpected artifacts though. e.g. 3 seconds hissing noise before dialogue. And music in background when I added (happy) in an attempt to control tone. Also don't understand how to control the S1 and S2 speakers...is it just random based on temp?

    > TODO Docker support

    Got this adapted pretty easily. Just latest nvidia cuda container, throw python and modules on it and change server to serve on 0.0.0.0. Does mean it pulls the model every time on startup though which isn't ideal

    • dragonwriter 13 minutes ago
      > Also don't understand how to control the S1 and S2 speakers...

      Do a clip with the speakers you want as the audio prompt, add the text of that clip (with speaker tags) of the clip at the beginning of your text prompt, and it clones the voices from your audio prompt for the output.

    • toebee 1 hour ago
      Thank you for the kind words! Dia wasn’t fine tuned on certain speaker, so you will get random voices every time you run it, unless you add a prompt / fix the seed.

      The outputs are a bit unstable, might need to add cleaner training data and run longer training sessions. Hopefully we can do something like OAI Whisper and update with better performing checkpoints!

    • yjftsjthsd-h 2 hours ago
      > Does mean it pulls the model every time on startup though which isn't ideal

      Surely it just downloads to a directory that can be volume mapped?

      • Havoc 2 hours ago
        Yep. I just didn't spend the time to track down the location tbh. Plus huggingface usually does links to a cache folder that I don't recall the location of

        Literally got cuda containers working earlier today so haven't spent a huge amount of time figuring things out

  • toebee 7 hours ago
    Hey HN! We’re Toby and Jay, creators of Dia. Dia is 1.6B parameter open-weights model that generates dialogue directly from a transcript.

    Unlike TTS models that generate each speaker turn and stitch them together, Dia generates the entire conversation in a single pass. This makes it faster, more natural, and easier to use for dialogue generation.

    It also supports audio prompts — you can condition the output on a specific voice/emotion and it will continue in that style.

    Demo page comparing it to ElevenLabs and Sesame-1B https://yummy-fir-7a4.notion.site/dia

    We started this project after falling in love with NotebookLM’s podcast feature. But over time, the voices and content started to feel repetitive. We tried to replicate the podcast-feel with APIs but it did not sound like human conversations.

    So we decided to train a model ourselves. We had no prior experience with speech models and had to learn everything from scratch — from large-scale training, to audio tokenization. It took us a bit over 3 months.

    Our work is heavily inspired by SoundStorm and Parakeet. We plan to release a lightweight technical report to share what we learned and accelerate research.

    We’d love to hear what you think! We are a tiny team, so open source contributions are extra-welcomed. Please feel free to check out the code, and share any thoughts or suggestions with us.

    • dangoodmanUT 1 hour ago
      I know it’s taboo to ask, but I must: where’s the dataset from? Very eager to play around with audio models myself, but I find existing datasets limiting
    • gfaure 6 hours ago
      Amazing that you developed this over the course of three months! Can you drop any insight into how you pulled together the audio data?
      • isoprophlex 5 hours ago
        +1 to this, amazing how you managed to deliver this, and iff you're willing to share i'd be most interested in learning what you did in terms of train data..!
    • heystefan 3 hours ago
      Could one usecase be generating an audiobook with this from existing books? I wonder if I could fine-tune the "characters" that speak these lines since you said it's a single pass whole the whole convo. Wonder if that's a limitation for this kind of a usecase (where speed is not imperative).
    • smusamashah 1 hour ago
      Hi! This is awesome for size and quality. I want to see a book reading example or try it myself.

      This is a tangent point but it would have been nicer if it wasn't a notion site. You could put the same page on github pages and it will be much lighter to open, navigate and link (like people trying to link some audio)

    • llm_nerd 2 hours ago
      This is a pretty incredible three month creation for a couple of people who had no experience with speech models.
    • nickthegreek 5 hours ago
      Are there any examples of the audio differences between the this and the larger model?
      • toebee 6 minutes ago
        We're still experimenting, so do not have samples yet from the larger model. All we have is Dia-1.6B at the moment.
    • bzuker 3 hours ago
      hey, this looks (or rather, sounds) amazing! Does it work with different languages or is it English only?
    • new_user_final 6 hours ago
      Easily 10 times better than recent OpenAI voice model. I don't like robotic voices.

      Example voices seems like over loud, over excitement like Andrew Tate, Speed or advertisement. It's lacking calm, normal conversation or normal podcast like interaction.

  • strobe 7 hours ago
    just in case, another opensource project using same name https://wiki.gnome.org/Apps/Dia/

    https://gitlab.gnome.org/GNOME/dia

    • toebee 7 hours ago
      Thanks for the heads-up! We weren’t aware of the GNOME Dia project. Since we focus on speech AI, we’ll make sure to clarify that distinction.
      • aclark 6 hours ago
        Ditto this! Dia diagram tool user here just noticing the name clash. Good luck with your Dia!! Assuming both can exist in harmony. :-)
        • mrandish 3 hours ago
          > Assuming both can exist in harmony.

          I'm sure they can... talk it over.

          I'll show myself out.

    • freedomben 5 hours ago
      Fun, I can't get to it because I can't get past the "Making sure you're not a bot!" page. It's just stuck at "calculating...". I understand the desire to slow down AI bots, but . If all the gnome apps are now behind this, they just completely shut down a small-time contributor. I love to play with Gnome apps and help out with things here and there, but I'm not going to fight with this damn thing to do so.
    • SoKamil 5 hours ago
      And another one, not open source but in AI sphere: https://www.diabrowser.com/
    • Magma7404 6 hours ago
      I know it's a bit ridiculous to see that as some kind of conspiracy, but I have seen a very long list of AI-related projects that got the same name as a famous open-source project, as if they wanted to hijack the popularity of those projects, and Dia is yet another example. It was relatively famous a few years ago and you cannot have forgotten it if you used Linux for more than a few weeks. It's almost done on purpose.
      • teddyh 5 hours ago
        The generous interpretation is that the AI hype people just didn’t know about those other projects, i.e. that they are neither open source developers, nor users.
        • gapan 4 hours ago
          Of course, how could they have known? Doing a basic web search before deciding on a name is so last year.
  • dindindin 2 hours ago
    Was this trained on Planet Money / NPR podcasts? The last audio (continuation of prompt) sounds eerily like Planet Money, I had to double check if my Spotify had accidentally started playing.
    • jelling 1 hour ago
      NPR voice is a thing.

      It started with Ira Glass voice and now the default voice is someone that sounds like they're not certain they should be saying the very banal thing they are about to say, followed by a hand-shake protocol of nervous laughter.

  • Versipelle 6 hours ago
    This is really impressive; we're getting close to a dream of mine: the ability to generate proper audiobooks from EPUBs. Not just a robotic single voice for everything, but different, consistent voices for each protagonist, with the LLM analyzing the text to guess which voice to use and add an appropriate tone, much like a voice actor would do.

    I've tried "EPUB to audiobook" tools, but they are really miles behind what a real narrator accomplishes and make the audiobook impossible to engage with

    • azinman2 4 hours ago
      Wouldn’t it be more desirable to hear an actual human on an audiobook? Ideally the author?
      • ks2048 54 minutes ago
        With 1M+ new books every year, that’s not possible for all but the few most popular.
      • Versipelle 3 hours ago
        > Wouldn’t it be more desirable to hear an actual human on an audiobook? Ideally the author?

        Of course, but it's not always available.

        For example, I would love an audiobook for Stanisław Lem's "The Invincible," as I just finished its video game adaptation, yet it simply doesn't exist in my native language.

        It's quite seldom that the author narrates the audiobooks I listen to, and sometimes the narrator does a horrible job, butchering the characters with exaggerated tones.

      • senordevnyc 3 hours ago
        Honestly, I’d say that’s true only for the author. Anyone else is just going to be interpreting the words to understand how to best convey the character / emotion / situation / etc., just like an AI will have to do. If an AI can do that more effectively than a human, why not?

        The author could be better, because they at least have other info beyond the text to rely on, they can go off-script or add little details, etc.

        • DrSiemer 3 hours ago
          As somebody who has listened to hundreds of audiobooks, I can tell you authors are generally not the best choice to voice their own work. They may know every intent, but they are writers, not actors.

          The most skilled readers will make you want to read books _just because they narrated them_. They add a unique quality to the story, that you do not get from reading yourself or from watching a video adaptation.

          Currently I'm in The Age of Madness, read by Steven Pacey. He's fantastic. The late Roy Dotrice is worth a mention as well, for voicing Game of Thrones and claiming the Guinness world record for most distinct voices (224) in one series.

          It will be awesome if we can create readings automatically, but it will be a while before TTS can compete with the best readers out there.

          • azinman2 1 hour ago
            I’d suggest even if the TTS sounded good, I’d still rather a human because:

            1. It’s a job that seems worthwhile to support, especially as it’s “practice” that only adds to a lifetime of work and improves their central skill set

            2. A voice actor will bring their own flare, just like any actor does to their job

            3. They (should) prepare for the book, understanding what it’s about in its entirety, and bring that context to the reading

    • mclau157 4 hours ago
      Realistic voice acting for audio books, realistic images for each page, realistic videos for each page, oh wait I just created a movie, maybe I can change the plot? Oh wait I just created a video game
  • xbmcuser 5 hours ago
    Wow first time I have felt that this could be the end of voice acting/audio book narration etc. The speed with with the ways things are changing how soon before you can make any book any novel into a complete audio video / movie or tv show.
  • enodios 1 hour ago
    The audio quality is seriously impressive. Any plans to add word-level timing maps? For my usecase that is a requirement, so unfortunately I cannot use this yet, but I would very much like to.
  • oehtXRwMkIs 1 hour ago
    Any plans for AMD GPU support? Maybe I'm missing something, but it's not working out of the box on a 7900xtx.
  • codingmoh 4 hours ago
    Hey, this is really cool! Curious how good the multi-language support is. Also - pretty wild that you trained the whole thing yourselves, especially without prior experience in speech models.

    Might actually be helpful for others if you ever feel like documenting how you got started and what the process looked like. I’ve never worked with TTS models myself, and honestly wouldn’t know where to begin. Either way, awesome work. Big respect.

  • dangoodmanUT 1 hour ago
    Has the same issue of cutting off the end of the provided text that many other models have.
  • sarangzambare 7 hours ago
    Impressive demo! We'd love to use this at https://useponder.ai

    time to first audio is something that is crucial for us to reduce the latency - wondering if dia works with output streaming?

    the python code snippet seems to imply that the entire audio bytes are generated directly?

    • toebee 6 hours ago
      Sounds awesome! I think it won't be very hard to run it using output streaming, although that might require beefier GPUs. Give us an email and we can talk more - nari.ai.contact at gmail dot com.

      It's way past bedtime where I live, so will be able to get back to you after a few hours. Thanks for the interest :)

  • instagary 1 hour ago
    Does this use the the mimi codec by moshi? If so it would be straighforward to get Dia running on iOS!
    • toebee 1 hour ago
      We use descript audio codec! I’m not sure if DAC works on iOS…
  • a2128 5 hours ago
    What's the training process like? I have some data in my language I'd love to use train it on my language seeing as it's English-only
  • IshKebab 7 hours ago
    Why does it say "join waitlist" if it's already available?

    Also, you don't need to explicitly create and activate a venv if you're using uv - it deals with that nonsense itself. Just `uv sync`.

    • toebee 7 hours ago
      We're envisioning a platform with a social aspect, so that is the biggest difference. Also, bigger models!

      We are aware of the fact that you do not need to create a venv when using pre-existing uv. Just added it for people spinning up new GPUs on cloud. But I'll update the README to make that a bit clearer. Thanks for the feedback :)

    • flakiness 7 hours ago
      Seek back a few tens of bytes which states "Play with a larger version of Dia"
  • eob 4 hours ago
    Bravo -- this is fantastic.

    I've been waiting for this ever since reading some interview with Orson Scott Card ages ago. It turns out he thinks of his novels as radio theater, not books. Which is a very different way to experience the audio.

  • 999900000999 5 hours ago
    Does this only support English?

    I would absolutely love something like this for practicing Chinese, or even just adding Chinese dialogue to a project.

    • buttercrab 44 minutes ago
      Hi! I'm Dia's developer. We currently only support English.
  • toebee 6 hours ago
    It is way past bedtime here, will be getting back to comments after a few hours of sleep! Thanks for all the kind words and feedback
  • isoprophlex 5 hours ago
    Incredible quality demo samples, well done. How's the performance for multilingual generation?
  • ivape 7 hours ago
    Darn, don't have the appropriate hardware.

    The full version of Dia requires around 10GB of VRAM to run.

    If you have a 16gb of VRAM, I guess you could pair this with a 3B param model along side it, or really probably only 1B param with reasonable context window.

    • toebee 6 hours ago
      We will work on a quantized version of the model, so hopefully you will be able to run it soon!

      We've seen Bark from Suno go from 16GB requirement -> 4GB requirement + running on CPUs. Won't be too hard, just need some time to work on it.

      • ivape 6 hours ago
        No doubt, these TTS models locally are what I'm looking for because I'm so done typing and reading :)
  • noiv 3 hours ago
    The demo page does fancy stuff when marking text and hitting cmd-d to create a bookmark :)
  • youssefabdelm 5 hours ago
    Anyone know if possible to fine-tune for cloning my voice?
  • verghese 5 hours ago
    How does this compare with Spark TTS?

    https://github.com/SparkAudio/Spark-TTS

  • popalchemist 5 hours ago
    This looks excellent, thank you for releasing openly.
  • vagabund 3 hours ago
    The huggingface spaces link doesn't work, fyi.

    Sounds awesome in the demo page though.

    • toebee 49 minutes ago
      We are in the progress of fixing it! Thanks for letting us know :)
  • stuartjohnson12 7 hours ago
    Impressive project! We'd love to use something like this over at Delfa (https://delfa.ai). How does this hold up from the perspective of stability? I've spoken to various folks working on voice models, and one thing that has consistently held Eleven Labs ahead of the pack from my experience is that their models seem to mostly avoid (while albeit not being immune to) accent shifts and distortions when confronted with unfamiliar medical terminology.

    A high quality, affordable TTS model that can consistently nail medical terminology while maintaining an American accent has been frustratingly elusive.

    • toebee 7 hours ago
      Interesting. I haven't thought of that problem before. I'm guessing a large enough audio dataset for medical terminology does not exist publicly.

      But AFAIK, even if you have just a few hours of audio containing specific terminology (and correct pronunciation), fine-tuning on that data will significantly improve performance.

  • pzo 6 hours ago
    Sounds great. Hope more language support in the future. In comparison Sesame CSM-1B sounds like trained on stoned people.
  • brumar 6 hours ago
    Impressive! Is it english only at the moment?
    • toebee 6 hours ago
      Unfortunately yes at the moment
  • hiAndrewQuinn 2 hours ago
    Is this English-only? I'm looking for a local model for Finnish dialogue to run.
  • film42 6 hours ago
    Very very impressive.
  • lostmsu 5 hours ago
    Does this only work for two voices? Can I generate an entire conversation between multiple people? Like this HN thread.
  • xhkkffbf 3 hours ago
    Are there different voices? Or only [s1] and [s2] in the examples?
  • zhyder 4 hours ago
    V v cool: first time I've seen such expressiveness in TTS for laughs, coughs, yelling about a fire, etc!

    What're the recommended GPU cloud providers for using such open-weights models?

  • xienze 6 hours ago
    How do you declare which voice should be used for a particular speaker? And can it created a cloned speaker voice from a sample?
    • toebee 6 hours ago
      You can add an audio prompt and prepend text corresponding to it in the script. You can get a feel for it by trying the second example in the Gradio interface!
  • jokethrowaway 3 hours ago
    Looking forward to try. My current go-to solution is E5-F2 (great cloning, decent delivery, ok audio quality, a lot of incoherence here and there forcing you to do multiple generations).

    I've just been massively disappointed by Sesame's CSM: on their gradio on the website it was generating flawless dialogs with amazing voice cloning. When running it local the voice cloning performance is awful.

  • user_4028b09 1 hour ago
    [dead]
  • mclau157 4 hours ago
    Will you support the other side with AI voice detection software to detect and block malicious voice snippets?