Rohan Chaubey

Gan.AI TTS Model & API Playground - First TTS model to support all 22 Indic languages + English

by

We are excited to launch Myna-mini TTS Model & API for a research preview. This is the first in the world to support high-fidelity, text-to-speech in all 22 official Indic Languages & English with code-mixing capabilities and free playground access.

Add a comment

Replies

Best
Suvrat Bhooshan
Hey Product Hunt! 👋 I'm Suvrat, Founder and CEO of Gan.AI. We're excited to launch the research preview of Myna-mini, our groundbreaking text-to-speech model and API. Our team brings together deep learning expertise from Stanford, FAIR, IITs, BITS, Samsung, Microsoft, and Adobe Research, with over 1000+ combined citations. 🌏 Myna-mini is the first TTS model to support all 22 Official Indic languages alongside English. We're bridging the language gap and bringing voice technology to over 1 billion people. Key features: 22 Indic languages + English 🗣️ 5 diverse voices from different regions of India 🎙️ Native code-mixed support 🔄 High-quality, natural-sounding voices 🔊 Easy-to-use API 🖥️ Free Playground access 🆓 Code-mixed support means Myna-mini can seamlessly handle text that mixes multiple languages in a single sentence or paragraph. This reflects how people naturally communicate, switching between languages mid-conversation. This research preview has some known issues, and we're eager for community feedback to shape its development. Tell us what you want to see in future iterations! 📊 🔍 Try it out now at playground.gan.ai! 🚀 Coming soon: Myna and Myna-Large: Set to be the world's largest TTS models in terms of both training data and parameter count Zero-shot cross-lingual voice cloning 🔁 Lip-sync API 🎬 Avatar API 📱 Our technology is trusted by world-renowned brands for personalizing video and audio campaigns, including Uber, Salesforce, Coca-Cola, Pepsi, Nestlé, the San Antonio Spurs, the Indian National Congress, and Four Seasons. 🏢 Our mission at Gan.AI is to make AI accessible and useful for everyone, regardless of the language they speak. Myna-mini is a big step towards that goal. 🎯 We'd love for you to give it a spin and share your thoughts. Whether you're building apps, creating content, or exploring new ways to communicate, Myna-mini is here to give voice to your ideas – in multiple languages, seamlessly mixed. 💡 Questions? Feedback? I'll be here all day to discuss. Let's explore how Myna-mini can empower your projects and how we can improve it further! 📈 Shoutout to @rohanrecommends for hunting us!
Rose Kamal Love
@rohanrecommends @suvrat This honestly looks like something we wanna use at kroto.one Have already shared with the CTO let's see what the tech genius thinks
Mohit Kinra
Congrats on the launch, team! Super excited to test it out :)
Ishita Jain
@kinraw Thanks for the support! The free Playground access is our way of letting everyone experience Myna-mini’s capabilities. Dive in and let us know how it can fit into your projects!
Suvrat Bhooshan
@kinraw Thank you! Eagerly awaiting feedback!
Ema Elisi
Excited to see Myna-mini launch, @suvrat! The support for 22 Indic languages is a game changer for TTS. Can't wait to test the code-mixing capabilities! 🎉 #PH #Makers
Ishita Jain
@ema_elisi Thanks Ema! Try typing in the script of the language for best results!
Saloni Jaju
@ema_elisi Thank you for the shoutout! Do check out the playground and tell us what you think :)
Kyrylo Silin
Hey Suvrat, I'm wondering about the potential applications. Do you envision this being used primarily for accessibility, content creation, or something else? How does the quality compare to other TTS models for languages like Hindi or Tamil? Congrats! :)
Ishita Jain
@kyrylosilin Hey Kyrylo, Here are some Use Cases for Myna-mini : 🎥 Content Creation: Make videos or podcasts in any language you need. ☎️ Customer Support: Talk to your customers in their language, literally. 📚 Education & Entertainment: Create characters or lessons that chat just like your audience. Do let us know how you finally use it!
Suvrat Bhooshan
@kyrylosilin We'll be releasing a technical report comparing the naturalness of our models with the rest of the world? Which Hindi/Tamil TTS models would you like us to benchmark against?
Harshita Jain
Congrats on the launch, Suvrat & team!! Super exciting!
Ishita Jain
@harshitajyn Thank you! Excited to hear about how you’re planning to use Myna-mini! It's designed to integrate easily, so you can start building multilingual experiences right away. Let us know how it goes
Suvrat Bhooshan
@harshitajyn THank you!!
Chuhaihao
Sounds impressive,
Saloni Jaju
@chuhaihao Thank you for the shoutout! Check out the playground and tell us what you think!
Michael Green
Congrats on the launch, Suvrat! Myna-mini sounds incredible, especially with its support for code-mixing in Indic languages. I'm curious about how the model handles different dialects—are there specific accents or regional variations included? Additionally, can you share more about the training data used for high-fidelity voice quality? This could provide insights into its performance. It's exciting to see Myna-mini bridging communication gaps for so many people. I'll definitely check out the free playground! Also, can't wait to see the future iterations with Myna and Myna-Large. Do you have any benchmarks or KPIs on expected improvements for those models? Thanks for the insights, and looking forward to testing it out!
Suvrat Bhooshan
@michaelgreen Hi Michael! We have worked with a diverse set of partners to source a wide variety of accents across the entire country in our training data. We will release a technical report along with the launch of our large models! Expect improvements around latency, TTFB, naturalness, and new metrics and evals!
Michael Green
Exciting launch, Suvrat! Myna-mini is a game changer for TTS, especially with its support for all 22 Indic languages. The code-mixed feature is a must-have given India's linguistic diversity. Can't wait to see how it evolves and the impact it will have on accessibility in voice tech. Upvoted! 🚀
Suvrat Bhooshan
@michaelgreen Thanks Michael! We wanted a multilingual model from the beginning to allow code mixing, and cross lingual voice cloning which is coming soon!
Elke
Congrats on the launch, Suvrat! The Myna-mini sounds incredible, especially the support for all 22 Indic languages and code-mixing—this is a game changer for many developers and content creators. I have a couple of questions: How do you plan to handle accents and dialects within those languages? And what’s the best way for developers to integrate the API into existing workflows? Excited to try it out!
Suvrat Bhooshan
@elke_qin Thank you! We'll be adding granular accent control depending on what you need! For now, you can type directly in the regional style, and it should be able to pick it up. API Integration is straightforward, you can generate the API key via the dashboard. Refer to our documentation for a detailed guide: https://developer.gan.ai/ Please let me know if you have any questions!
charles shiro
Congrats on the launch, Suvrat! 🌟 Myna-mini sounds like a game changer in the TTS space, especially with its support for all 22 official Indic languages. Bridging that language gap is no small feat! I love that you’re incorporating code-mixing; it's about time TTS reflects real-world communication.
Suvrat Bhooshan
@charles_ I knowww! Thank you for your support!
Majid Izadi
Congratulation on the launch, great product, good job by you and your team. any plan on adding similar new languages like Persian or Arabic? also, may I know what is the technical process for adding new languages? what is the input, how long does it take? overall great product on the edge of tech, good luck today.
Suvrat Bhooshan
@m4jiz Thanks Majid! We have to source training data for each language, and vet it manually. It takes some time. Input is the text that you want to generate, in the native script of the language. We are able to achieve close to RTF of 1.0!
Kady ouilina
The focus on privacy and data security is reassuring. It's so important for companies to prioritize this, especially with advanced AI technologies. Keep it up!
Suvrat Bhooshan
@kadyouilina We appreciate your recognition of our privacy and security efforts. It's central to our work. We prioritize local processing, data anonymization, regular audits, and transparency. Your feedback helps us maintain high standards.
Ira M. Cassidy
It is doing amazing work in conversational AI! I can't wait to see how the Myna-large model will perform in real-world applications.
Suvrat Bhooshan
@ira_m_cassidy Fingers Crossed!!
Sagar Kava
Congratulations on the launch @suvrat ! This is fantastic news for the AI and TTS community. Supporting all 22 Indic languages alongside English is a huge achievement and can help make technology accessible to over a billion people. Is there detailed documentation and sample code available to help developers get started with the API quickly?
Suvrat Bhooshan
@sagar_kava Thanks! You can find it here: https://developer.gan.ai/
Ditarth Desai
Congratulations @Suvrat on your successful launch.
Suvrat Bhooshan
@ditarth_wbs Thank you!!
Stuti Pareek
Absolutely blown away by the best-in-class multilingual code-mixed TTS! The support for Indian languages & seamless language blending is incredibly useful. 🔥
Ishita Jain
@stuti_pareek1 Thank you! Do keep a look out for our upcoming models for even more applications: Myna and Myna-Large: Largest TTS models, currently in training Zero-shot cross-lingual voice cloning Lip-sync API Avatar API
Eric@VMEG
Yes! Making AI accessible and useful for everyone, regardless of the language they speak,which is the most inspiring words I have learned. A lot of people would benefit from Gan.AI. Cheers!
Ishita Jain
@eric_sung918 Thank you for the support! Eagerly awaiting your feedback!
Jenifer Lamberto
This tech sounds promising, especially the cross-lingual voice cloning. I’m curious how well it works across different languages. Great to see research pushing boundaries
Suvrat Bhooshan
@jenifer_lamberto Please try it out, even the current speakers can speak all 23 languages for you to get a sense of!
Avikalp Gupta
It was nice to play around with the tool, thanks for including the playground. Just a quick suggestion, the example you have in the intro video is called "Code Switching" not "Code mixing". This would be an example of code-mixing: Yo bro, क्या कर रहा है कल शाम को? मेरे घर पर midnight को एक house party है।We are going to "जै माता दी let's rock" it! When I tried the above sentence, it didn't do as poorly as I expected it to. It was actually pretty workable. I can imagine using this tool for sound generation without a sound studio and then using my own audio editing tools to make it work for my videos. I would suggest adding some way to indicate tonality. I tried code mixing without transliteration (just to test it out - it is often to handle such cases). I see that the team is yet to implement that. I am really looking forward to trying out (and helping test out) the future versions of the tool. All the very best! This is amazing.
Suvrat Bhooshan
@avikalp_gupta Thank you for the kind words, Avikalp! We use "Code Switching", and "Code mixing" interchangeably in our messaging, and were loose with definitions internally. 😅 We're working on adding more control features to the audio generation pipeline! The next generation of models should support both without the need of transliteration as well! Would love to learn more about how you foresee using this, and features we can prioritize for you!
Mitia
Well done guys! Congratulation on the launch and big support from our entire team!