Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

Drop 5/14: Introducing Bulbul V3, our latest text-to-speech model. It raises the bar for how human it sounds, while being super robust. In an independent third-party human listening study, Bulbul V3 delivers the highest listener preference, and low error rates across use-cases and languages. See details in our blog,... show more

Pratyush Kumar

33,240 subscribers

424,383 görüntüleme • 7 ay önce •via X (Twitter)

Eğitim Haberler & Politika Bilim & Teknoloji

Anya Rossi• Live Now

Private livecam show

38 Yorum

Pratyush Kumar profil fotoğrafı

Pratyush Kumar7 ay önce

In a blind study conducted by @JoshTalksLive, listeners compared Bulbul V3, ElevenLabs (v3 alpha and v2.5 flash), and Cartesia Sonic-3 with over 20,000 votes. Bulbul V3 tops the scores for 8kHz audio, setting a new benchmark for speech synthesis for voice agents.

Pratyush Kumar profil fotoğrafı

Pratyush Kumar7 ay önce

The listeners also tagged real failure modes allowing us to evaluate stability. Bulbul V3 comes out on top, with the lowest average error rates.

Pratyush Kumar profil fotoğrafı

Pratyush Kumar7 ay önce

We also evaluated for the long-tail of language challenges such as speaking numerics, technical content, and named entities. Bulbul V3 consistently has the lowest error rates across languages.

Pratyush Kumar profil fotoğrafı

Pratyush Kumar7 ay önce

Bulbul V3 is live. This is your month to go all in. Unlimited usage open through February. The mic is yours!

India in Pixels by Ashris 🔱 profil fotoğrafı

India in Pixels by Ashris 🔱7 ay önce

Incredible work, team Sarvam! Proud of you all!

akash profil fotoğrafı

akash7 ay önce

who ever makes these on brand world class animations 🫡

ezioAuditoreeeee profil fotoğrafı

ezioAuditoreeeee7 ay önce

Insaneee one this is... My Grandfather is kinda shocked at how good this turned out to be. He's a die hard fan of Ponniyin Selvan. Who let TTS Cook...🤯🤯

Ankit Khandelwal profil fotoğrafı

Ankit Khandelwal7 ay önce

Deep Tech is funny. Everyone thinks you're useless until the one day you drop something great and everyone loses their mind. Keeping morale high until that point is crucial.

Anshuman Mahapatra profil fotoğrafı

Anshuman Mahapatra7 ay önce

Excited for drop 14/14👀

Samuel Cardillo profil fotoğrafı

Samuel Cardillo7 ay önce

in hebrew it means penis v3

Rasputin profil fotoğrafı

Rasputin7 ay önce

Another one

Vansh Thakur profil fotoğrafı

Vansh Thakur7 ay önce

If Sarvam creates a big LLM chatbot like ChatGPT, Gemini, Perplexity then rest all these Sarvam models will automatically get a huge network effect boost.

Ella Ke profil fotoğrafı

Ella Ke7 ay önce

אני בהלם שקראתם לזה בולבול

Ankit Khandelwal profil fotoğrafı

Ankit Khandelwal7 ay önce

It is an inflection point where the world realizes that India-based deep learning labs can also deliver. This will change investors' views, and they will start taking riskier bets on Indian deep tech startups more confidently.

Ersan - beyninikullan.com profil fotoğrafı

Ersan - beyninikullan.com7 ay önce

Bülbül

akash profil fotoğrafı

akash7 ay önce

lesssgooo bulbul

dattasai profil fotoğrafı

dattasai7 ay önce

Amazing work guys!! We got TTS..! Sarvam Vision is also really impressive..!

Himanshu Kumar profil fotoğrafı

Himanshu Kumar7 ay önce

@pratykumar, that listener preference data is a strong indicator of Bulbul V3's enhanced human-like quality.

NS profil fotoğrafı

NS7 ay önce

Good going @pratykumar.

मनीष तिवारी profil fotoğrafı

मनीष तिवारी7 ay önce

ratan ji ki awaj me

UP 10T Economic Goal™ profil fotoğrafı

UP 10T Economic Goal™7 ay önce

Congratulations to the Sarvam Team for amazing work.

Vansh Thakur profil fotoğrafı

Vansh Thakur7 ay önce

Pratyush, I request you to make a single website for web users and an app for mobile users in which a user can easily navigate all 14 models of yours in a single place.

Gaurav Tewatia profil fotoğrafı

Gaurav Tewatia7 ay önce

I would suggest that u partner with indian influencers like podcasters or something so that they can automatically translate their entire podcast into other languages for greater visibility

Arjun Singh profil fotoğrafı

Arjun Singh7 ay önce

Damn 🔥🔥

🇮🇳 𐄳 इन्द्रजाल 𐄳 profil fotoğrafı

🇮🇳 𐄳 इन्द्रजाल 𐄳7 ay önce

One suggestion: can we add a pure Hindi mode? Don’t want to deal with that Hinglish crap.

Sanskar Pandey profil fotoğrafı

Sanskar Pandey7 ay önce

amazing work!

Priyankar Pal profil fotoğrafı

Priyankar Pal7 ay önce

Nice name

Niteesh Yadav ( नितीश ) profil fotoğrafı

Niteesh Yadav ( नितीश )7 ay önce

Huge kudos for building something that’s not just strong on the tech, but also so thoughtfully and tastefully designed.

MS profil fotoğrafı

MS7 ay önce

Can da model handle “Hinglish” ? That’s imp given our desi urban lifestyle …-also ur models can be used for Filmy world in subtitles n post prod edits etc ..that’s a niche market in itself

Yadu A profil fotoğrafı

Yadu A7 ay önce

Why no pure #Marathi language option given in text to speech @SarvamAI ?? Also need more voice options as many sound robotic. Need smooth and sobre voices, not blunt, harsh robotic ones.

KCAerospace🚀🚀 profil fotoğrafı

KCAerospace🚀🚀7 ay önce

YOOO this is so cool So glad to see Indian AI labs getting SOTA

Nio profil fotoğrafı

Nio7 ay önce

Great job guys i will add it to my movizai tool

Ashish Dogra profil fotoğrafı

Ashish Dogra7 ay önce

you guys are killing it

Sanat Kumar Mahapatra profil fotoğrafı

Sanat Kumar Mahapatra7 ay önce

@manojzxc

Krrish Agarwalla profil fotoğrafı

Krrish Agarwalla7 ay önce

Hi Pratyush! Why don't you build a full duplex model ? It is will more natural

TheRaven profil fotoğrafı

TheRaven7 ay önce

Huge leap 👏 If Bulbul V3 truly combines more human-like prosody + low error rates across languages, that’s a big milestone especially for multilingual markets like India.

Lokesh KR profil fotoğrafı

Lokesh KR7 ay önce

🔥🔥🔥

Yash profil fotoğrafı

Yash7 ay önce

Awesome job ! Just tried Telugu, its nice. Still Sounds a slightly robtic though, a bit like old time radio . Still better than V2. 👍 do you have plans to add intention/emotion style to the tone in the future releases so that it sounds more natural?

Benzer Videolar

Introducing Saaras V4, our most capable speech to text model yet. It delivers strong performance across both English and Indian languages, with better accuracy over a much wider range of speech. Saaras V4 was built with specific attention to noise, accents, dialects and mixed-language speech. Read more in the blog:

Introducing Saaras V4, our most capable speech to text model yet. It delivers strong performance across both English and Indian languages, with better accuracy over a much wider range of speech. Saaras V4 was built with specific attention to noise, accents, dialects and mixed-language speech. Read more in the blog:

Sarvam

44,151 görüntüleme • 3 gün önce

Meet the all-new Bulbul 🚀 Natural, familiar speech in 11 Indian languages, with authentic accents that sound just like India. ▪️Not robotic. Not rehearsed. Just real. ▪️Lightning-fast performance ▪️Custom voices for your brand From lower latency and India-first pricing to wider language, Bulbul sets a new benchmark for Speech AI in India. If you’re building for India, say it with Bulbul! Watch it in action ↓

Sarvam

274,327 görüntüleme • 1 yıl önce

Introducing Scribe — the most accurate Speech to Text model. It has the highest accuracy on benchmarks, outperforming previous state-of-the-art models such as Gemini 2.0 and OpenAI Whisper v3. It’s now the leading model for English, Spanish, Italian, and many more. With support for 99 languages, speaker diarization, character-level timestamps, and non-speech events such as laughing.

Introducing Scribe — the most accurate Speech to Text model. It has the highest accuracy on benchmarks, outperforming previous state-of-the-art models such as Gemini 2.0 and OpenAI Whisper v3. It’s now the leading model for English, Spanish, Italian, and many more. With support for 99 languages, speaker diarization, character-level timestamps, and non-speech events such as laughing.

ElevenLabs

465,139 görüntüleme • 1 yıl önce

Introducing Monsoon ASR ⚡️ Voice Arena Speech recognition does not have a model problem anymore. It has a data problem. The best ASR systems are approaching human-level performance in English. But move into the long tail of the world's languages, especially real, conversational speech and error rates can still be 5-10X higher. Today, we're releasing Monsoon ASR: a new generation of training data built specifically to close that gap. 50 languages. 100,000+ hours. Dense spontaneous speech. And one goal: Single-digit WER across the world's languages.🧵

Introducing Monsoon ASR ⚡️ Voice Arena Speech recognition does not have a model problem anymore. It has a data problem. The best ASR systems are approaching human-level performance in English. But move into the long tail of the world's languages, especially real, conversational speech and error rates can still be 5-10X higher. Today, we're releasing Monsoon ASR: a new generation of training data built specifically to close that gap. 50 languages. 100,000+ hours. Dense spontaneous speech. And one goal: Single-digit WER across the world's languages.🧵

Shobhit Banga

461,665 görüntüleme • 14 saat önce

Introducing Gemini 3.1 Flash TTS 🗣️, our latest text to speech model with scene direction, speaker level specificity, audio tags, more natural + expressive voices, and support for 70 different languages. Available via our new audio playground in AI Studio and in the Gemini API!

Introducing Gemini 3.1 Flash TTS 🗣️, our latest text to speech model with scene direction, speaker level specificity, audio tags, more natural + expressive voices, and support for 70 different languages. Available via our new audio playground in AI Studio and in the Gemini API!

Logan Kilpatrick

805,885 görüntüleme • 5 ay önce

Introducing Lightning V3 - it beats every model we tested against. ElevenLabs, Cartesia, OpenAI. Lightning sets a new SOTA with V3 in conversational text-to-speech. → Highest MOS score for conversational TTS at 3.9 → ~76% win rate vs gpt-4o-mini-tts on naturalness → 15 languages with mid-sentence code-switching → Built from scratch for voice agents, not read-aloud Every TTS model sounds clean in a demo. You type a sentence and you get beautiful audio. Voice agents don't work that way. They stream. They're generating audio in real-time chunks with half the context missing. That's where everything breaks. A great reading voice and a great conversational voice are fundamentally different things. A conversational voice has to sound like it's thinking - with the pauses, the rhythm shifts, the reactions. It has to handle the way real people actually talk, including switching languages mid-sentence. That's what V3 does. V3.1 also ships voice cloning. 5 to 15 seconds of audio, no fine-tuning, production-grade clone across 15 languages. Blog link in the comments.

Introducing Lightning V3 - it beats every model we tested against. ElevenLabs, Cartesia, OpenAI. Lightning sets a new SOTA with V3 in conversational text-to-speech. → Highest MOS score for conversational TTS at 3.9 → ~76% win rate vs gpt-4o-mini-tts on naturalness → 15 languages with mid-sentence code-switching → Built from scratch for voice agents, not read-aloud Every TTS model sounds clean in a demo. You type a sentence and you get beautiful audio. Voice agents don't work that way. They stream. They're generating audio in real-time chunks with half the context missing. That's where everything breaks. A great reading voice and a great conversational voice are fundamentally different things. A conversational voice has to sound like it's thinking - with the pauses, the rhythm shifts, the reactions. It has to handle the way real people actually talk, including switching languages mid-sentence. That's what V3 does. V3.1 also ships voice cloning. 5 to 15 seconds of audio, no fine-tuning, production-grade clone across 15 languages. Blog link in the comments.

Sudarshan Kamath

71,298 görüntüleme • 6 ay önce

Today, we’re excited to introduce Miso One, the most emotive voice model in the world. Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency. We’ve open-sourced the model weights, with API access coming soon. Hear how Miso One sounds in the thread below.

Today, we’re excited to introduce Miso One, the most emotive voice model in the world. Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency. We’ve open-sourced the model weights, with API access coming soon. Hear how Miso One sounds in the thread below.

Aoden Teo

5,269,129 görüntüleme • 3 ay önce

Today, we’re launching Bland Speech v3 - The world's first Human Speech Engine. In Design Arena’s Audio Realism benchmark, Speech v3 is the top model, outranking Elevenlabs, Grok, Cartesia, and OpenAI. Trained on over 100 million real, human conversations: businesses, developers, and creators can now create any audio they can imagine. To showcase this, we used 5 seconds of old footage to help James, a 49 year old father who recently suffered a stroke, get his voice back. Try it for free at Full length documentary below🧵

Today, we’re launching Bland Speech v3 - The world's first Human Speech Engine. In Design Arena’s Audio Realism benchmark, Speech v3 is the top model, outranking Elevenlabs, Grok, Cartesia, and OpenAI. Trained on over 100 million real, human conversations: businesses, developers, and creators can now create any audio they can imagine. To showcase this, we used 5 seconds of old footage to help James, a 49 year old father who recently suffered a stroke, get his voice back. Try it for free at Full length documentary below🧵

Bland

2,131,924 görüntüleme • 1 ay önce

Today, we released Gemini 3.5 Live Translate, our latest audio model for live speech-to-speech translation. It supports over 70 languages and starts translating as soon as you start talking, streaming translations while listening to what you say next. No awkward pauses or choppy audio, just real connection without language barriers. So, how does it work? 🤔 The model is able to make split-second decisions to juggle speed and translation quality so conversations actually feel fluid, human, and natural. In order to do this, the model must receive and contextualize the input while simultaneously outputting the translated speech. Through this process, Gemini 3.5 Live Translate manages to stay mere seconds behind each speaker and can even maintain pacing, pitch, and intonation across extended sessions. See it in action below, or try it yourself in the Google Translate app on iOS & Android.

Today, we released Gemini 3.5 Live Translate, our latest audio model for live speech-to-speech translation. It supports over 70 languages and starts translating as soon as you start talking, streaming translations while listening to what you say next. No awkward pauses or choppy audio, just real connection without language barriers. So, how does it work? 🤔 The model is able to make split-second decisions to juggle speed and translation quality so conversations actually feel fluid, human, and natural. In order to do this, the model must receive and contextualize the input while simultaneously outputting the translated speech. Through this process, Gemini 3.5 Live Translate manages to stay mere seconds behind each speaker and can even maintain pacing, pitch, and intonation across extended sessions. See it in action below, or try it yourself in the Google Translate app on iOS & Android.

Google AI

3,989,279 görüntüleme • 3 ay önce

Speech translation has been one of the longest-running ML efforts at Google, and we’ve come a long way. Gemini 3.5 Live Translate is our latest speech-to-speech model, supporting 70+ languages. It enables more natural conversations across languages in everyday products and apps. Here’s an example of how partners at Grab are helping connect travelers with drivers. 🚗 Rolling out in Google Translate and via the Live API in Google AI Studio.

Speech translation has been one of the longest-running ML efforts at Google, and we’ve come a long way. Gemini 3.5 Live Translate is our latest speech-to-speech model, supporting 70+ languages. It enables more natural conversations across languages in everyday products and apps. Here’s an example of how partners at Grab are helping connect travelers with drivers. 🚗 Rolling out in Google Translate and via the Live API in Google AI Studio.

Jeff Dean

35,828 görüntüleme • 3 ay önce

Meet Hibiki, our simultaneous speech-to-speech translation model, currently supporting 🇫🇷➡️🇬🇧. Hibiki produces spoken and text translations of the input speech in real-time, while preserving the speaker’s voice and optimally adapting its pace based on the semantic content of the source speech. Based on objective and human evaluations, Hibiki outperforms previous systems for quality, naturalness and speaker similarity and approaches human interpreters. 🧵

Meet Hibiki, our simultaneous speech-to-speech translation model, currently supporting 🇫🇷➡️🇬🇧. Hibiki produces spoken and text translations of the input speech in real-time, while preserving the speaker’s voice and optimally adapting its pace based on the semantic content of the source speech. Based on objective and human evaluations, Hibiki outperforms previous systems for quality, naturalness and speaker similarity and approaches human interpreters. 🧵

kyutai

167,717 görüntüleme • 1 yıl önce

introducing Gemini 3.5 Transcribe: our most precise speech-to-text model yet, designed for intelligent voice interactions this new model: - seamlessly handles self-corrections - removes filler words to deliver clean, formatted text - understands your natural intent and speaking style - captures audio, even in noisy environments - supports 85+ languages, including regional accents and diverse dialects try it today in the Gemini API and in AI Studio

introducing Gemini 3.5 Transcribe: our most precise speech-to-text model yet, designed for intelligent voice interactions this new model: - seamlessly handles self-corrections - removes filler words to deliver clean, formatted text - understands your natural intent and speaking style - captures audio, even in noisy environments - supports 85+ languages, including regional accents and diverse dialects try it today in the Gemini API and in AI Studio

Google AI Studio

238,290 görüntüleme • 1 ay önce

OpenAI's S2S preview is polished but it still thinks in steps. Speech → text → model → text → speech. That's not how humans converse. Introducing Hydra. A native speech-to-speech model that doesn't wait for turn-taking, doesn't flatten emotion into text, and doesn't break when you interrupt it mid-sentence. Hydra reasons asynchronously, speaks and listens simultaneously, and preserves emotion because it never leaves the audio domain. It's still in beta, but the shift is obvious. If you want early access, the link is in the comments. Here's a preview of what that looks like -

OpenAI's S2S preview is polished but it still thinks in steps. Speech → text → model → text → speech. That's not how humans converse. Introducing Hydra. A native speech-to-speech model that doesn't wait for turn-taking, doesn't flatten emotion into text, and doesn't break when you interrupt it mid-sentence. Hydra reasons asynchronously, speaks and listens simultaneously, and preserves emotion because it never leaves the audio domain. It's still in beta, but the shift is obvious. If you want early access, the link is in the comments. Here's a preview of what that looks like -

Sudarshan Kamath

329,378 görüntüleme • 6 ay önce

We're launching Veena TTS 🪕 on June 20 Our flagship text-to-speech model for Indian languages 🇮🇳 Natural, expressive, and actually sounds like us. We’re launching two models: Veena Lite >Open-source and lightweight >4 unique, natural-sounding voices >The first open-source TTS model designed for real creative use cases Veena Max >Available via our web app >15 expressive, high-quality voices >Perfect for storytelling, dubbing, voiceovers, and content creation

We're launching Veena TTS 🪕 on June 20 Our flagship text-to-speech model for Indian languages 🇮🇳 Natural, expressive, and actually sounds like us. We’re launching two models: Veena Lite >Open-source and lightweight >4 unique, natural-sounding voices >The first open-source TTS model designed for real creative use cases Veena Max >Available via our web app >15 expressive, high-quality voices >Perfect for storytelling, dubbing, voiceovers, and content creation

Dheemanth Reddy

80,431 görüntüleme • 1 yıl önce

You imagine it. We create it. 🎨 Introducing #PixelStudio, a first-of-its-kind image generator on #Pixel9 powered by an on-device diffusion model¹ running on Tensor G4 and our Imagen 3 text-to-image model in the cloud.

You imagine it. We create it. 🎨 Introducing #PixelStudio, a first-of-its-kind image generator on #Pixel9 powered by an on-device diffusion model¹ running on Tensor G4 and our Imagen 3 text-to-image model in the cloud.

Made by Google

28,283 görüntüleme • 2 yıl önce

Virgin Galactic is revolutionizing the operating model and economics of human spaceflight. To improve the clarity and quality of information available to our large retail investor base, we made a video presenting our multi-year growth plan and business model. We encourage our investors and fans to watch and hear how we are creating value now, while laying the foundation for substantial growth in the years to come!

Virgin Galactic is revolutionizing the operating model and economics of human spaceflight. To improve the clarity and quality of information available to our large retail investor base, we made a video presenting our multi-year growth plan and business model. We encourage our investors and fans to watch and hear how we are creating value now, while laying the foundation for substantial growth in the years to come!

Virgin Galactic

32,113 görüntüleme • 2 yıl önce

Introducing Scribe v2 Realtime – the most accurate real-time Speech to Text model. Built for voice agents, meeting notetakers, and live applications, it transcribes in 150ms across 90+ languages, including English, French, German, Italian, Spanish, Portuguese, Hindi, and Japanese. Available today by API and through ElevenLabs Agents.

Introducing Scribe v2 Realtime – the most accurate real-time Speech to Text model. Built for voice agents, meeting notetakers, and live applications, it transcribes in 150ms across 90+ languages, including English, French, German, Italian, Spanish, Portuguese, Hindi, and Japanese. Available today by API and through ElevenLabs Agents.

ElevenLabs

318,433 görüntüleme • 10 ay önce

In Love With An Object Why do people name their cars, talk to their plants, or feel emotionally attached to places and objects? The human brain is wired to see agency and personality everywhere. In extreme cases this can even turn into romantic attachment to objects—a phenomenon psychologists call objectophilia. It sounds strange, but it reveals something fascinating about how our minds create meaning and connection.

In Love With An Object Why do people name their cars, talk to their plants, or feel emotionally attached to places and objects? The human brain is wired to see agency and personality everywhere. In extreme cases this can even turn into romantic attachment to objects—a phenomenon psychologists call objectophilia. It sounds strange, but it reveals something fascinating about how our minds create meaning and connection.

Kate Shelor

10,761 görüntüleme • 6 ay önce