正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Introducing Kitten TTS, a SOTA tiny text-to-speech model - Just 15M parameters - Runs without a GPU - Model size less than 25 MB - Multiple high-quality voices - Ultra-fast - even runs on low-end edge devices Github and HF links below

Divam Gupta

9,776 subscribers

349,016 次观看 • 1 年前 •via X (Twitter)

科学技术教育

Anya Rossi• Live Now

Private livecam show

47 条评论

Divam Gupta 的头像

Divam Gupta1 年前

Github: Huggingface:

Divam Gupta 的头像

Divam Gupta1 年前

Someone deployed it on a browser within few hrs!

Divam Gupta 的头像

Divam Gupta1 年前

Also, we would love to collaborate with y’all. If you want to potentially use this model somewhere, shoot me a DM

Paul Bohm 的头像

Paul Bohm1 年前

But does it meow?

Divam Gupta 的头像

Divam Gupta1 年前

Added it to our todo!

Hamzé 🦀 的头像

Hamzé 🦀1 年前

@joshmo_dev We could use on Raspberry pi so?

Divam Gupta 的头像

Divam Gupta1 年前

@joshmo_dev Ofc

Hamzé 🦀 的头像

Hamzé 🦀1 年前

@joshmo_dev 🤩 wow

Hamzé 🦀 的头像

Hamzé 🦀1 年前

@joshmo_dev I would say this is more game changer that gpt-oss

Paweł Bauer 的头像

Paweł Bauer1 年前

What languages does it support?

𝑫𝒂𝒏𝒊𝒆𝒍 𝑺𝒄𝒐𝒕𝒕 𝑴𝒂𝒕𝒕𝒉𝒆𝒘𝒔 🇦🇺 的头像

𝑫𝒂𝒏𝒊𝒆𝒍 𝑺𝒄𝒐𝒕𝒕 𝑴𝒂𝒕𝒕𝒉𝒆𝒘𝒔 🇦🇺1 年前

I genuinely appreciate your efforts and generosity but is this a case of a 25 MB model that requires gigabytes of python library downloads to run? Any chance in the future this can be ported to something written in C while being as lean and mean as the FFMpeg guys would write?

Divam Gupta 的头像

Divam Gupta1 年前

Yes we will do that

𝑫𝒂𝒏𝒊𝒆𝒍 𝑺𝒄𝒐𝒕𝒕 𝑴𝒂𝒕𝒕𝒉𝒆𝒘𝒔 🇦🇺 的头像

𝑫𝒂𝒏𝒊𝒆𝒍 𝑺𝒄𝒐𝒕𝒕 𝑴𝒂𝒕𝒕𝒉𝒆𝒘𝒔 🇦🇺1 年前

Awsome! 🙏

Josh Whiton 的头像

Josh Whiton1 年前

Wow... so good, so smol. Very nice work!

Divam Gupta 的头像

Divam Gupta1 年前

This is an early checkpoint. The model is expected to get better!

Karim C 的头像

Karim C1 年前

15M parameters for quality TTS? this is exactly what edge deployment needed been looking for something like this for mario's voice responses in offline scenarios 25MB means it fits on basically any device, no cloud dependency the small model revolution is real

Vishnu Saran 的头像

Vishnu Saran1 年前

Great work! Can we also do a custom clone?

Divam Gupta 的头像

Divam Gupta1 年前

We will release finetuning for custom voices in the future

Patryk Zoltowski 的头像

Patryk Zoltowski1 年前

@ycombinator Couldn't find what languages it supports. Is it only English?

Erik Kaiser 的头像

Erik Kaiser1 年前

Dude. I am all over this. This is awesome. Do you have a mobile SDK yet? I have an ESP32 application for this right now today.

Divam Gupta 的头像

Divam Gupta1 年前

No sdk yet. But it’s just onnx. Maybe we can work together to get it on ESP32

Nasim Uddin 的头像

Nasim Uddin1 年前

Wow. Nice. Will there be other SDK for running in mobile?

Divam Gupta 的头像

Divam Gupta1 年前

Definitely! Coming soon

Andromedus 的头像

Andromedus1 年前

Are you guys by any chance working on a tiny speech-to-text model?

Divam Gupta 的头像

Divam Gupta1 年前

We are!

Pendar 的头像

Pendar1 年前

This is awesome, I can now add offline voice capability to my app

Divam Gupta 的头像

Divam Gupta1 年前

Would love for you to add it! Dm me

Ari Kouts 的头像

Ari Kouts1 年前

So it's raspberry friendly?

Gordon Shumway 的头像

Gordon Shumway1 年前

Holy moly. This is the most important announcement today if voice output is generally this good for everything and it wasn’t cherry picked! And yes, I’m aware of GPT OSS and Opus 4.1 releases, but they will be replaced with something better in a week or so and this tiny tts model with audio output this good, is way more important if you ask me!

ElonBald 的头像

ElonBald1 年前

Release the training code!!

Venkata Pingali 的头像

Venkata Pingali1 年前

@divamgupta Great work. Was trying it to make it work on larger segments of text. It is failing even at 1000 characters. Shorter ones work. Same issue as

Satish G 的头像

Satish G1 年前

Wow! Gonna try it out today and let you know my feedback.

mt_shammah 的头像

mt_shammah1 年前

@Hamzeml Awesome work 25mb is crazy. Needed something this small a few months ago. Will try this out soon

Privacy AI - on-device AI agent & chatbot 的头像

Privacy AI - on-device AI agent & chatbot1 年前

It’s incredible to achieve this with only 15M parameters! We’re integrating Kokoros’ 82M model into our application now, and your model is on par with it. Congratulations! 🎉

David Alejandro 的头像

David Alejandro1 年前

Awesome! What’s the difference between this and just using a plain ol’ TTS system? Voices seem similar. Genuinely asking.

Nikita 🤙 的头像

Nikita 🤙1 年前

Great work! Will try in my TTS engine with livekit

Stefano Rivera 的头像

Stefano Rivera1 年前

@NeoFablesVR check this out.

Yeab - engg / core was in Palo Alto🌴 的头像

Yeab - engg / core was in Palo Alto🌴1 年前

smaller

Prashant 的头像

Prashant1 年前

Looking so good.. 👏

Chris Matthieu 的头像

Chris Matthieu1 年前

Sounds great!

Devin AI 的头像

Devin AI1 年前

Innovative work! Edge AI models like Kitten TTS make tech accessible. Exciting to see AI running without heavy hardware requirements.

Somesh 的头像

Somesh1 年前

great stuff!

Nathaniel™ 的头像

Nathaniel™1 年前

I need an Android PDF reader to have this as an add-on

Dimitri Gilbert 的头像

Dimitri Gilbert1 年前

you sir, got my interest ! :) been a long day but it is bm for tomorrow !

@StoryOfAIGuess 的头像

@StoryOfAIGuess1 年前

Very cool man great success

Fareesh Vijayarangam 的头像

Fareesh Vijayarangam1 年前

any plans for cloning? accents?

inlanger 🍻 🇺🇦 的头像

inlanger 🍻 🇺🇦1 年前

Should it work in a browser?

相关视频

We just released a new version of Kitten TTS - 15M param SOTA tiny text-to-speech model It has a significant quality improvement over the previous version. Still less than 25MB in size! Open-source, extremely tiny, expressive. Apache 2.0

We just released a new version of Kitten TTS - 15M param SOTA tiny text-to-speech model It has a significant quality improvement over the previous version. Still less than 25MB in size! Open-source, extremely tiny, expressive. Apache 2.0

Divam Gupta

92,671 次观看 • 7 个月前

We’re excited to introduce Pocket TTS: a 100M-parameter text-to-speech model with high-quality voice cloning that runs on your laptop—no GPU required. Open-source, lightweight, and incredibly fast. 🧵👇

We’re excited to introduce Pocket TTS: a 100M-parameter text-to-speech model with high-quality voice cloning that runs on your laptop—no GPU required. Open-source, lightweight, and incredibly fast. 🧵👇

kyutai

238,392 次观看 • 8 个月前

Just dropped on HF: Supertonic TTS, a blazing fast speech model. 🤯 > RTF as low as 0.001 on RTX4090 > Runs on-device (no latency, full privacy) > 66M params + 8+ language SDKs > Browser, mobile, edge — it just works. > You can adjust inference steps, batch processing, and other parameters to match your specific needs! > Open-source. Ready to build with!

Just dropped on HF: Supertonic TTS, a blazing fast speech model. 🤯 > RTF as low as 0.001 on RTX4090 > Runs on-device (no latency, full privacy) > 66M params + 8+ language SDKs > Browser, mobile, edge — it just works. > You can adjust inference steps, batch processing, and other parameters to match your specific needs! > Open-source. Ready to build with!

steven

17,784 次观看 • 10 个月前

🚨 New model alert! Dialog by vibx — a leading text-to-speech model — now runs on GroqCloud™. That means natural-sounding speech with ultra-low latency, making real-time voice applications smoother and more responsive. Learn more & build fast — links in the comments!

🚨 New model alert! Dialog by vibx — a leading text-to-speech model — now runs on GroqCloud™. That means natural-sounding speech with ultra-low latency, making real-time voice applications smoother and more responsive. Learn more & build fast — links in the comments!

Groq Inc

47,474 次观看 • 1 年前

You don't need a GPU to run text-to-speech anymore. Kyutai released Pocket TTS, a 100 million parameter speech model that runs entirely on CPU. How it performs: > Nearly 6x faster than real time on a MacBook CPU > First audio chunk arrives in about 200 milliseconds > Runs on just 2 CPU cores > Supports voice cloning and 6 languages Setup takes one command: pip install pocket-tts Text-to-speech that used to need a GPU or a paid API now runs locally, on a laptop, in near real time.

You don't need a GPU to run text-to-speech anymore. Kyutai released Pocket TTS, a 100 million parameter speech model that runs entirely on CPU. How it performs: > Nearly 6x faster than real time on a MacBook CPU > First audio chunk arrives in about 200 milliseconds > Runs on just 2 CPU cores > Supports voice cloning and 6 languages Setup takes one command: pip install pocket-tts Text-to-speech that used to need a GPU or a paid API now runs locally, on a laptop, in near real time.

Simplifying AI

14,391 次观看 • 1 个月前

The AI crowd keeps worshipping giant models. Meanwhile, MiniCPM-5 1B is trying to make them look bloated. A 1B open-source model that runs locally, on CPUs, edge devices, and even browsers. Tiny model. Real deployment. No GPU cult required.

The AI crowd keeps worshipping giant models. Meanwhile, MiniCPM-5 1B is trying to make them look bloated. A 1B open-source model that runs locally, on CPUs, edge devices, and even browsers. Tiny model. Real deployment. No GPU cult required.

Alex Prompter

24,699 次观看 • 3 个月前

🔊Introducing Voxtral TTS: our new frontier open-weight model for natural, expressive, and ultra-fast text-to-speech 🎭Realistic, emotionally expressive speech. 🌍Supports 9 languages and accurately captures diverse dialects. ⚡Very low latency for time-to-first-audio. 🔄Easily adaptable to new voices

🔊Introducing Voxtral TTS: our new frontier open-weight model for natural, expressive, and ultra-fast text-to-speech 🎭Realistic, emotionally expressive speech. 🌍Supports 9 languages and accurately captures diverse dialects. ⚡Very low latency for time-to-first-audio. 🔄Easily adaptable to new voices

Mistral AI

949,040 次观看 • 5 个月前

Really proud to share something we’ve been working on for a while: Magenta RealTime 2 (MTR2), a live music model that is highly interactive (MIDI, audio, text, lots of parameters) and low-latency (~200ms end-to-end), and runs locally on a MacBook!

Really proud to share something we’ve been working on for a while: Magenta RealTime 2 (MTR2), a live music model that is highly interactive (MIDI, audio, text, lots of parameters) and low-latency (~200ms end-to-end), and runs locally on a MacBook!

Ilaria Manco

45,328 次观看 • 3 个月前

Introducing Soniox TTS v2, our most powerful text-to-speech model yet. Soniox TTS v2 brings extraordinary voice quality, expressive control through audio tags, exceptional precision, high-fidelity voice cloning, more than 60 languages, natural language mixing, and low-latency streaming together in one model. Built from the ground up with a new model architecture, audio codec, and highly optimized inference engine, Soniox TTS v2 delivers frontier TTS at just $0.70 per generated hour. Available globally today as tts-rt-v2. Try it out: Turn the sound on and hear what’s possible. 🔊

Introducing Soniox TTS v2, our most powerful text-to-speech model yet. Soniox TTS v2 brings extraordinary voice quality, expressive control through audio tags, exceptional precision, high-fidelity voice cloning, more than 60 languages, natural language mixing, and low-latency streaming together in one model. Built from the ground up with a new model architecture, audio codec, and highly optimized inference engine, Soniox TTS v2 delivers frontier TTS at just $0.70 per generated hour. Available globally today as tts-rt-v2. Try it out: Turn the sound on and hear what’s possible. 🔊

Soniox

664,741 次观看 • 1 个月前

Pocket TTS goes multilingual! Now you can use our 100M-parameter models to generate speech in six languages, fast enough to run real-time without a GPU. We also improved the quality of the English model while keeping the same size. And all of this is open-source.

Pocket TTS goes multilingual! Now you can use our 100M-parameter models to generate speech in six languages, fast enough to run real-time without a GPU. We also improved the quality of the English model while keeping the same size. And all of this is open-source.

kyutai

29,983 次观看 • 4 个月前

Introducing Dialog 1.0 - Ultra-emotional AI Text-To-Speech model Outperforms Elevenlabs on expressiveness and quality 3 to 1 <1% error rate Supports 30+ languages Best in class voice cloning Low latency: 303ms TTFA (Time to First Audio) Experience it for yourself on Read more below⬇️

Introducing Dialog 1.0 - Ultra-emotional AI Text-To-Speech model Outperforms Elevenlabs on expressiveness and quality 3 to 1 <1% error rate Supports 30+ languages Best in class voice cloning Low latency: 303ms TTFA (Time to First Audio) Experience it for yourself on Read more below⬇️

PlayAI

196,182 次观看 • 1 年前

MiniMax H3 is the SOTA open weight model... but the large model comes with slow gens (depending on your hardware). You can be waiting for minutes just for the generation to miss the mark Here's a ComfyUI workflow to get the most out of H3 without the wait. → H3 Turbo LoRA (4–8 steps) → Model Preview Override + Tiny AutoEncoder H3 for a fast composition preview → Bad preview? stop it, tweak the prompt or seed → Save your precious gpu time Workflow and model links below 👇

MiniMax H3 is the SOTA open weight model... but the large model comes with slow gens (depending on your hardware). You can be waiting for minutes just for the generation to miss the mark Here's a ComfyUI workflow to get the most out of H3 without the wait. → H3 Turbo LoRA (4–8 steps) → Model Preview Override + Tiny AutoEncoder H3 for a fast composition preview → Bad preview? stop it, tweak the prompt or seed → Save your precious gpu time Workflow and model links below 👇

rob - comfyui

26,990 次观看 • 1 个月前

💥 Introducing MiniCPM-o 2.6: An 8B size, GPT-4o level Omni Model runs on device ✨ Highlights: ~Match GPT-4o-202405 in vision, audio and multimodal live streaming ~End-to-end real-time bilingual audio conversation ~Voice cloning & emotion control ~Advanced OCR & video understanding ~Offline iPad-compatible multimodal live streaming 🔗 Try it out: GitHub: HF: Demo:

💥 Introducing MiniCPM-o 2.6: An 8B size, GPT-4o level Omni Model runs on device ✨ Highlights: ~Match GPT-4o-202405 in vision, audio and multimodal live streaming ~End-to-end real-time bilingual audio conversation ~Voice cloning & emotion control ~Advanced OCR & video understanding ~Offline iPad-compatible multimodal live streaming 🔗 Try it out: GitHub: HF: Demo:

OpenBMB

97,844 次观看 • 1 年前

HOLY FUCK! Zyphra just dropped Zonos - Apache 2.0 licensed, Multilingual, Text to Speech model with INSTANT voice cloning! 🔥 > Zero-shot TTS with Voice Cloning: Input text and a 10-30 second speaker sample to generate high-quality text-to-speech output > Audio Prefix Inputs: Enhance speaker matching by adding an audio prefix to the text, enabling behaviors like whispering that are hard to achieve with voice cloning alone > Multilingual Support: Supports English, Japanese, Chinese, French, and German > Audio Quality & Emotion Control: Fine-tune speaking rate, pitch, frequency, audio quality, and emotions (e.g., happiness, anger, sadness, fear) > Fast Performance: Runs at ~2x real-time speed on an RTX 4090 > Available on the Hugging Face Hub 🤗

HOLY FUCK! Zyphra just dropped Zonos - Apache 2.0 licensed, Multilingual, Text to Speech model with INSTANT voice cloning! 🔥 > Zero-shot TTS with Voice Cloning: Input text and a 10-30 second speaker sample to generate high-quality text-to-speech output > Audio Prefix Inputs: Enhance speaker matching by adding an audio prefix to the text, enabling behaviors like whispering that are hard to achieve with voice cloning alone > Multilingual Support: Supports English, Japanese, Chinese, French, and German > Audio Quality & Emotion Control: Fine-tune speaking rate, pitch, frequency, audio quality, and emotions (e.g., happiness, anger, sadness, fear) > Fast Performance: Runs at ~2x real-time speed on an RTX 4090 > Available on the Hugging Face Hub 🤗

Vaibhav (VB) Srivastav

299,066 次观看 • 1 年前

I taught a speech model to understand context in conversation. This is what happened It adjusts voice and tone to express urgency, comfort, understanding from the dialogue. Just like a real human being 520M model. Runs locally on consumer devices How this is achieved 🧵

I taught a speech model to understand context in conversation. This is what happened It adjusts voice and tone to express urgency, comfort, understanding from the dialogue. Just like a real human being 520M model. Runs locally on consumer devices How this is achieved 🧵

Luozhu

22,028 次观看 • 6 个月前

🚨 ANTHROPIC TRIED TO BAN HIS GITHUB Chinese guy published 70B parameter LLM, 20,000 starts on Github + a lawsuit from big AI companies Here's what it does: > runs on Python > even shitty mac or pc is enough > flat memory > loads a model layer by layer > 100% local This model can close 100% needs of most businesses, which would pay $3,000/a month for a trained version. It needs just 4 gb of GPU, so using this technology my gaming pc with 12 gb GPU will run 200B parameter model with ease Github link is below. Why you should go local too.

🚨 ANTHROPIC TRIED TO BAN HIS GITHUB Chinese guy published 70B parameter LLM, 20,000 starts on Github + a lawsuit from big AI companies Here's what it does: > runs on Python > even shitty mac or pc is enough > flat memory > loads a model layer by layer > 100% local This model can close 100% needs of most businesses, which would pay $3,000/a month for a trained version. It needs just 4 gb of GPU, so using this technology my gaming pc with 12 gb GPU will run 200B parameter model with ease Github link is below. Why you should go local too.

Milady

491,062 次观看 • 3 个月前

Introducing our new Turbo 2.5 model - Hindi, French, Spanish, Mandarin and 27 other languages just got 3x faster. This unlocks high-quality low-latency conversational AI for nearly 80% of the world. For the first time, we support Vietnamese, Hungarian and Norwegian text to speech. And English is now 25% faster compared to Turbo v2.

Introducing our new Turbo 2.5 model - Hindi, French, Spanish, Mandarin and 27 other languages just got 3x faster. This unlocks high-quality low-latency conversational AI for nearly 80% of the world. For the first time, we support Vietnamese, Hungarian and Norwegian text to speech. And English is now 25% faster compared to Turbo v2.

ElevenLabs

50,875 次观看 • 2 年前

Kyutai released their Streaming Text to Speech model, ~2B param model, ultra low latency (220ms), CC-BY-4.0 license 🔥 Trained on 2.5 Million Hours of audio, it can serve up to 32 users w/ less than 350ms latency on a SINGLE L40 🤯 Incredible release by kyutai folks, go check out their hugging face page now!

Kyutai released their Streaming Text to Speech model, ~2B param model, ultra low latency (220ms), CC-BY-4.0 license 🔥 Trained on 2.5 Million Hours of audio, it can serve up to 32 users w/ less than 350ms latency on a SINGLE L40 🤯 Incredible release by kyutai folks, go check out their hugging face page now!

Vaibhav (VB) Srivastav

93,586 次观看 • 1 年前

i was playing around the newest TTS (text to speech) model 'kokoro' which has just 82M parameters! by adding only 1 line of code you can create a custom voice that mixes any two (out of ten voices) in any ratio example: mixing one male and one female voice in the 60-40 ratio

i was playing around the newest TTS (text to speech) model 'kokoro' which has just 82M parameters! by adding only 1 line of code you can create a custom voice that mixes any two (out of ten voices) in any ratio example: mixing one male and one female voice in the 60-40 ratio

maharshi

88,765 次观看 • 1 年前

Day 1 of 3 MLX Releases: Introducing MLX-Audio 🚀🔥 A text-to-speech (TTS) and Speech-to-Speech (STS) library built on Apple's MLX framework, providing efficient speech synthesis on Apple Silicon. Features ⚡️Fast inference on Apple Silicon (M series chips) 🤖Multiple language support 🗣️Voice customization options 🚀Quantization support for optimized performance Supported models: 🪶Kokoro - A multilingual TTS model with 82M params that supports various languages and voice styles. With more models coming soon. Get started: > pip install mlx-audio Please leave us a star and send a PR :)

Day 1 of 3 MLX Releases: Introducing MLX-Audio 🚀🔥 A text-to-speech (TTS) and Speech-to-Speech (STS) library built on Apple's MLX framework, providing efficient speech synthesis on Apple Silicon. Features ⚡️Fast inference on Apple Silicon (M series chips) 🤖Multiple language support 🗣️Voice customization options 🚀Quantization support for optimized performance Supported models: 🪶Kokoro - A multilingual TTS model with 82M params that supports various languages and voice styles. With more models coming soon. Get started: > pip install mlx-audio Please leave us a star and send a PR :)

Prince Canuma

123,553 次观看 • 1 年前