Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

ReVISE: Self-Supervised Speech Resynthesis with Visual Input for Universal and Generalized Speech Enhancement abs: project page:

AK

516,345 subscribers

34,993 views • 3 years ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

6 Comments

Alex Volkov (Thursd/AI)3 years ago

This is so weird haah

RY⚡3 years ago

Wait.. is it..forming the speech just from the visual input? No, right?

GanLike3 years ago

😁🙀

Clément3 years ago

AHAHAHAHHAHAHHAHAHHA

Reyon Gifto Immanuel3 years ago

@scispace_ Summarize this

SciSpace3 years ago

@_akhaliq @ReyonGifto Here you go 👇 It proposes a unified model, ReVISE, for Generalized Speech Enhancement, with the goal of improving intelligibility, quality, and video synchronization. Results show that ReVISE greatly suppresses noise and improves quality, showing its applicability

Related Videos

📢New paper We are announcing ReVISE, the first universal audio-visual speech enhancement model powered by SSL. paper: demo: w/ Yossi Adi @TalRemez BowenShi Jacob Donley

📢New paper We are announcing ReVISE, the first universal audio-visual speech enhancement model powered by SSL. paper: demo: w/ Yossi Adi @TalRemez BowenShi Jacob Donley

Wei-Ning Hsu

46,382 views • 3 years ago

Mini-Omni 2 understands image, audio and text inputs all via end-to-end voice conversations with users 🔥 > Understands and processes images, speech, and text > Generates real-time speech responses > Supports interruptions during speech Technical Overview: > Concatenates image, audio, and text features for input. > Uses text-guided delayed parallel output for real-time speech > Involves encoder adaptation, modal alignment, and multimodal fine-tuning Best part: MIT licensed ⚡

Mini-Omni 2 understands image, audio and text inputs all via end-to-end voice conversations with users 🔥 > Understands and processes images, speech, and text > Generates real-time speech responses > Supports interruptions during speech Technical Overview: > Concatenates image, audio, and text features for input. > Uses text-guided delayed parallel output for real-time speech > Involves encoder adaptation, modal alignment, and multimodal fine-tuning Best part: MIT licensed ⚡

Vaibhav (VB) Srivastav

45,453 views • 1 year ago

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.

AK

290,517 views • 3 years ago

SceneScape: Text-Driven Consistent Scene Generation abs: project page: text-driven perpetual view generation -- synthesizing long videos of arbitrary scenes solely from an input text describing the scene and camera poses

SceneScape: Text-Driven Consistent Scene Generation abs: project page: text-driven perpetual view generation -- synthesizing long videos of arbitrary scenes solely from an input text describing the scene and camera poses

AK

73,258 views • 3 years ago

Open-source project Soundstorm (AI generated speech from Google Research) is going to give Elevenlabs a run for it's money: The text-to-speech project specializes in dialogue between multiple parties, and is available on Github:

Open-source project Soundstorm (AI generated speech from Google Research) is going to give Elevenlabs a run for it's money: The text-to-speech project specializes in dialogue between multiple parties, and is available on Github:

AI Breakfast

326,486 views • 3 years ago

Introducing... Amica! 👩‍🦰🤖 Amica is an open source interface for interactive communication with 3D characters with voice synthesis, speech recognition, visual understanding, and an emotion system. 🔗

Introducing... Amica! 👩‍🦰🤖 Amica is an open source interface for interactive communication with 3D characters with voice synthesis, speech recognition, visual understanding, and an emotion system. 🔗

Arbius

41,769 views • 2 years ago

🔥 Ming-UniAudio: The 「Nano Banana」moment for speech is here! A single model for universal understanding, generation & free-form editing. First Unified Continuous Tokenizer 「MingTok-Audio」and Unified Und & Gen Speech LLM built on it. First Universal Free-form Speech Editing Model Without Timestamp Condition. First FreeForm Audio Editing Benchmark. 💻 GitHub: 🤗 Tokenizer: 🤗 Model: base: edit: 🤗 Benchmark: 🌍 blog: #AI #Speech #SpeechLLM #LLM #GenerativeAI #Audio #ASR #TTS #SpeechEditing

🔥 Ming-UniAudio: The 「Nano Banana」moment for speech is here! A single model for universal understanding, generation & free-form editing. First Unified Continuous Tokenizer 「MingTok-Audio」and Unified Und & Gen Speech LLM built on it. First Universal Free-form Speech Editing Model Without Timestamp Condition. First FreeForm Audio Editing Benchmark. 💻 GitHub: 🤗 Tokenizer: 🤗 Model: base: edit: 🤗 Benchmark: 🌍 blog: #AI #Speech #SpeechLLM #LLM #GenerativeAI #Audio #ASR #TTS #SpeechEditing

Ant Ling

21,365 views • 11 months ago

Supervised learning has held 3D Vision back for too long. Meet RayZer — a self-supervised 3D model trained with zero 3D labels: ❌ No supervision of camera & geometry ✅ Just RGB images And the wild part? RayZer outperforms supervised methods (as 3D labels from COLMAP is noisy) 🌐 Project: (1/4)

Supervised learning has held 3D Vision back for too long. Meet RayZer — a self-supervised 3D model trained with zero 3D labels: ❌ No supervision of camera & geometry ✅ Just RGB images And the wild part? RayZer outperforms supervised methods (as 3D labels from COLMAP is noisy) 🌐 Project: (1/4)

Hanwen Jiang

71,182 views • 1 year ago

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

167,672 views • 1 year ago

Kyutai TTS and Unmute are now open source! The text-to-speech is natural, customizable, and fast: it can serve 32 users with a 350ms latency on a single L40S. Try it out and get started on the project page:

Kyutai TTS and Unmute are now open source! The text-to-speech is natural, customizable, and fast: it can serve 32 users with a 350ms latency on a single L40S. Try it out and get started on the project page:

kyutai

172,287 views • 1 year ago

You can use Cerebras' gpt-oss-120b to build realistic speech-to-speech voice interfaces with emotion via Hume AI's new EVI 3. Perfect for your next voice ai project! Link to try below 👇

You can use Cerebras' gpt-oss-120b to build realistic speech-to-speech voice interfaces with emotion via Hume AI's new EVI 3. Perfect for your next voice ai project! Link to try below 👇

Cerebras

22,034 views • 1 year ago

does your audio-input LLM secretly know how to generate speech? yes but it sounds like a scary demon! deepdream but for speech: starting with noise audio, I tried optimizing the probability of a desired transcription, while keeping the model frozen. volume up! 🎧

does your audio-input LLM secretly know how to generate speech? yes but it sounds like a scary demon! deepdream but for speech: starting with noise audio, I tried optimizing the probability of a desired transcription, while keeping the model frozen. volume up! 🎧

Matt Henderson

86,530 views • 1 month ago

One for all, and all for one 🧧 Introducing Ming-flash-omni-2.0: A specialist in every domain, unified as a capable generalist. A gift from Ling =) - Unified Acoustic Synthesis: Speech, audio, and music combined for unbounded creativity; - "Seeing" to "Knowing": Moving beyond input to true deep semantic understanding; - Native Visual Fusion: Seamless generation, editing, and segmentation;

One for all, and all for one 🧧 Introducing Ming-flash-omni-2.0: A specialist in every domain, unified as a capable generalist. A gift from Ling =) - Unified Acoustic Synthesis: Speech, audio, and music combined for unbounded creativity; - "Seeing" to "Knowing": Moving beyond input to true deep semantic understanding; - Native Visual Fusion: Seamless generation, editing, and segmentation;

Ant Ling

1,925,484 views • 7 months ago

Here’s a clip of Frisco Mayor Jeff Cheney announcing changes coming to the Citizen Input Process. And there’s also a clip of me ROASTING him for it. Only Tyrants are afraid of Free Speech. The First Amendment and Texas Open Meetings Act protect my speech here.

Here’s a clip of Frisco Mayor Jeff Cheney announcing changes coming to the Citizen Input Process. And there’s also a clip of me ROASTING him for it. Only Tyrants are afraid of Free Speech. The First Amendment and Texas Open Meetings Act protect my speech here.

Marc Palasciano

20,979 views • 6 months ago

Universal-Streaming is our real-time speech-to-text model — built for voice agents with ~300ms latency, high accuracy, unlimited concurrency, and just $0.15/hour. 👉 Test it live in the playground →

Universal-Streaming is our real-time speech-to-text model — built for voice agents with ~300ms latency, high accuracy, unlimited concurrency, and just $0.15/hour. 👉 Test it live in the playground →

AssemblyAI

1,017,880 views • 1 year ago

🚨 BREAKING: The self-proclaimed champion of "free speech" just called for Rachel Maddow and Nicolle Wallace to be fired—for using their free speech about the Cancer survivor at his Speech. These aren't the words of a president. They're the words of a dictator. And let’s not forget—this is the same guy slashing funding for cancer research.

🚨 BREAKING: The self-proclaimed champion of "free speech" just called for Rachel Maddow and Nicolle Wallace to be fired—for using their free speech about the Cancer survivor at his Speech. These aren't the words of a president. They're the words of a dictator. And let’s not forget—this is the same guy slashing funding for cancer research.

Brian Krassenstein

1,521,878 views • 1 year ago

The Realtime API is officially out of beta and ready for your production voice agents! We’re also introducing gpt-realtime—our most advanced speech-to-speech model yet—plus new voices and API capabilities: 🔌 Remote MCPs 🖼️ Image input 📞 SIP phone calling ♻️ Reusable prompts

The Realtime API is officially out of beta and ready for your production voice agents! We’re also introducing gpt-realtime—our most advanced speech-to-speech model yet—plus new voices and API capabilities: 🔌 Remote MCPs 🖼️ Image input 📞 SIP phone calling ♻️ Reusable prompts

OpenAI Developers

562,147 views • 1 year ago

What if voice AI could understand context, intent, and emotion — not just words? 🎙️ Meet Seed Speech 2.0. A new speech AI stack designed for natural, expressive conversations — with major upgrades in both text-to-speech (TTS) and speech recognition (ASR). Built for developers and creators who want natural conversational speech. What’s inside: - Natural and expressive speech generation - Prompt-controlled emotion and tone - High-accuracy multilingual recognition - Strong contextual reasoning Built for: 🎧 Voice content creation 🤖 AI assistants ☎️ Customer service 🎬 Dubbing & subtitling 📊 Audio-video analysis Voice AI built for expression, accuracy, and understanding. Try it now → Read more about Seed Speech here → #SeedSpeech #VoiceAI #SpeechAI #AI

What if voice AI could understand context, intent, and emotion — not just words? 🎙️ Meet Seed Speech 2.0. A new speech AI stack designed for natural, expressive conversations — with major upgrades in both text-to-speech (TTS) and speech recognition (ASR). Built for developers and creators who want natural conversational speech. What’s inside: - Natural and expressive speech generation - Prompt-controlled emotion and tone - High-accuracy multilingual recognition - Strong contextual reasoning Built for: 🎧 Voice content creation 🤖 AI assistants ☎️ Customer service 🎬 Dubbing & subtitling 📊 Audio-video analysis Voice AI built for expression, accuracy, and understanding. Try it now → Read more about Seed Speech here → #SeedSpeech #VoiceAI #SpeechAI #AI

BytePlus

19,840 views • 5 months ago

🏹 Britain can restore free speech with political will and legal reforms. Watch barrister Jon Holbrook's key findings in Prosperity Institute's latest paper, Reversing Britain's Free Speech Recession, and how to restore robust protections for free speech in the UK. Watch below

🏹 Britain can restore free speech with political will and legal reforms. Watch barrister Jon Holbrook's key findings in Prosperity Institute's latest paper, Reversing Britain's Free Speech Recession, and how to restore robust protections for free speech in the UK. Watch below

Prosperity Institute

1,235,139 views • 2 months ago