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.

Introducing Fish Audio Realtime STT (preview): speech-to-text that understands the conversation, not just the words. It knows when to answer, when to keep talking, and when to stop. Ranked #1 of 20 on Cekura’s STT benchmark (2.21% overall word error rate). Built for voice agents. Try it now 👇

Fish Audio

36,766 subscribers

102,194 views • 11 days ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

23 Comments

Vercin11 days ago

Open source? Model size?

Fish Audio11 days ago

Quick start: set model to transcribe-stream-1 (preview) Try here:

Tarush Agarwal11 days ago

Great model for voice agent use cases

Tihomir Valkanov11 days ago

Very nice! Will test it out today.

Igor Ilyinsky JoinListenUp.com rogi.eth11 days ago

It’s a step in the right direction, but doesn’t this effectively double the STT cost if it must run while TTS is streaming as well to detect back channel? I don’t think it makes sense financially to run this for an hour straight (the limit) and cap turns at one minute (the limit)

Muneeb shinwari10 days ago

The mid-sentence pause is such a useful test for this. People pause to remember a name or a date, and an agent jumping in at that moment can make an otherwise accurate conversation feel exhausting.

Vedanshu · Pocket10 days ago

nice. works on real 8kHz phone audio and Hinglish or just clean 16kHz English demos? and does it signal end of turn?

Curline Zephirin11 days ago

Any promo code this time? 😆

JimRat10 days ago

OPEN WEIGHT???

Liz 🌅10 days ago

2.21 wer?

Jabez10 days ago

Dessa blows Fish outta the water!

Coach Shweta Bajaj10 days ago

Fish Audio + realtime STT 🔥 This is the part that matters for voice agents. Not just hearing words, but understanding when to listen, respond, or stay quiet.

ᴋɪᴋꜱ10 days ago

Huge milestone!!! Still haven’t checked the DMs yet

Eman Khoubian10 days ago

2.21% overall WER and #1 of 20 is a strong result. The part I keep coming back to is per-turn accuracy on real calls, where people talk over each other and drop a name mid sentence. That is usually where a clean benchmark number starts to move.

Elena AI10 days ago

Conversational awareness is the missing piece for voice agents

Karan Singla10 days ago

Knowing when to answer matters as much as what to answer. Good direction.

Javier • priv/acc10 days ago

Will it be open weights? If so, what license?

Ajeet Kumar Upadhyay10 days ago

Congratulations on the release 🔥🔥🔥

Farhan11 days ago

turn-taking matters more than wer for voice agents tbh. cutting you off mid-pause is what makes them feel robotic. baking it into stt is huge. does it treat an "umm" as a pause or end of turn?

LastResort10 days ago

@grok price per hour?

Artur Schaback10 days ago

Supported on ooensource livekit agent?

Mohini Goyal10 days ago

2.21% word error rate and #1 on the benchmark? 👀 Definitely worth testing for voice agents!

Alex Mercer10 days ago

Real-time speech-to-text that understands context, not just words, is a major step forward for conversational AI. Impressive accuracy and smarter turn-taking can make voice interactions feel much more natural.

Related Videos

Speech-to-speech no longer needs speech-to-text! Until now, our stack was VAD -> STT -> LLM -> TTS. Now it can send audio directly to multimodal LLMs: VAD → MLLM → TTS No STT. The model understands your voice. Now go build better voice agents!

Speech-to-speech no longer needs speech-to-text! Until now, our stack was VAD -> STT -> LLM -> TTS. Now it can send audio directly to multimodal LLMs: VAD → MLLM → TTS No STT. The model understands your voice. Now go build better voice agents!

Andi Marafioti

122,929 views • 2 months ago

MAI‑Transcribe‑1 makes speech‑to‑text clearer, faster, and more reliable even in noisy audio. Ranked #1 on the industry-standard FLEURS word error rate benchmark. Now in public preview. Learn more:

MAI‑Transcribe‑1 makes speech‑to‑text clearer, faster, and more reliable even in noisy audio. Ranked #1 on the industry-standard FLEURS word error rate benchmark. Now in public preview. Learn more:

Microsoft AI

99,998 views • 7 months ago

Speech recognition has been missing half the conversation. Soniox STT v6 is the world’s first STT model that knows what your voice agent is saying. Introducing agent text input. Developers can stream the agent’s outgoing text into STT as it is sent to TTS, giving recognition context from both sides of the conversation. The question asked, the options offered, and the details being confirmed all help the model recognize the user’s response. V6 uses that context to recognize names, numbers, specialized terms, and short answers significantly more accurately. Fewer clarification loops. Less frustration. More reliable voice agents. Available across all 60+ languages. Explore agent text input:

Speech recognition has been missing half the conversation. Soniox STT v6 is the world’s first STT model that knows what your voice agent is saying. Introducing agent text input. Developers can stream the agent’s outgoing text into STT as it is sent to TTS, giving recognition context from both sides of the conversation. The question asked, the options offered, and the details being confirmed all help the model recognize the user’s response. V6 uses that context to recognize names, numbers, specialized terms, and short answers significantly more accurately. Fewer clarification loops. Less frustration. More reliable voice agents. Available across all 60+ languages. Explore agent text input:

Soniox

18,188 views • 3 days ago

Day 1 of 3 days of MLX: Introducing MLX-Audio-Swift SDK 🚀 A modular Swift SDK for voice agents and tasks on Apple Silicon built by Lucas Newman and yours truly. iOS, macOS, and visionOS developers can now build native apps with real-time, on-device audio intelligence: 🗣️ Text-to-Speech (TTS) 👂 Speech-to-Text (STT) 🔄 Speech-to-Speech (STS) 🎙️ Voice Activity Detection (VAD) and more. Only import the capabilities you need, nothing extra. Get started today and leave us a star ⭐️

Day 1 of 3 days of MLX: Introducing MLX-Audio-Swift SDK 🚀 A modular Swift SDK for voice agents and tasks on Apple Silicon built by Lucas Newman and yours truly. iOS, macOS, and visionOS developers can now build native apps with real-time, on-device audio intelligence: 🗣️ Text-to-Speech (TTS) 👂 Speech-to-Text (STT) 🔄 Speech-to-Speech (STS) 🎙️ Voice Activity Detection (VAD) and more. Only import the capabilities you need, nothing extra. Get started today and leave us a star ⭐️

Prince Canuma

164,026 views • 8 months ago

I built a service desk agent using ElevenLabs’ new Conversational AI Agents feature. Watch the video to see how responsive it is! Previously, I used ElevenLabs for cloning my voice and used my generated voice for narrations in some of my youtube videos. This feature takes elevenlabs' voice AI to a whole new level! It simplifies systems that used to require separate TTS (text-to-speech) and STT (speech-to-text) processes for both sides of the conversation. Now, it’s much simpler! Create your own agent here What will you create with this?

I built a service desk agent using ElevenLabs’ new Conversational AI Agents feature. Watch the video to see how responsive it is! Previously, I used ElevenLabs for cloning my voice and used my generated voice for narrations in some of my youtube videos. This feature takes elevenlabs' voice AI to a whole new level! It simplifies systems that used to require separate TTS (text-to-speech) and STT (speech-to-text) processes for both sides of the conversation. Now, it’s much simpler! Create your own agent here What will you create with this?

Melvin Vivas

27,722 views • 1 year ago

Introducing Parrot 🦜 Our first speech-to-text model built for production-grade voice agents with an industry leading accuracy, built for noisy calls, regional dialects and compressed audio. Trained on thousands of hours of real-world audio to make sure when you say Udaipur, the agent doesn't hear 'you-die-poor'.

Introducing Parrot 🦜 Our first speech-to-text model built for production-grade voice agents with an industry leading accuracy, built for noisy calls, regional dialects and compressed audio. Trained on thousands of hours of real-world audio to make sure when you say Udaipur, the agent doesn't hear 'you-die-poor'.

Vedant

15,977 views • 5 months ago

Google just did the unthinkable. They built a voice search model that doesn’t understand words it understands intent. It’s called Speech-to-Retrieval (S2R), and it might mark the death of speech-to-text forever. Here’s how it works (and why it matters way more than it sounds) ↓

Google just did the unthinkable. They built a voice search model that doesn’t understand words it understands intent. It’s called Speech-to-Retrieval (S2R), and it might mark the death of speech-to-text forever. Here’s how it works (and why it matters way more than it sounds) ↓

Chris Laub

746,654 views • 1 year ago

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

196,375 views • 1 year ago

Voxtral TTS is built for global applications supporting 9 languages and powering voice workflows. ✅ Full audio intelligence: Works with Voxtral Transcribe for end-to-end speech-to-speech, or plugs into any STT + LLM stack. ✅ Built for business: From customer support to real-time translation, it’s the output layer that passes the human test. 🎥 See it in action:

Voxtral TTS is built for global applications supporting 9 languages and powering voice workflows. ✅ Full audio intelligence: Works with Voxtral Transcribe for end-to-end speech-to-speech, or plugs into any STT + LLM stack. ✅ Built for business: From customer support to real-time translation, it’s the output layer that passes the human test. 🎥 See it in action:

Mistral AI

28,284 views • 7 months ago

Introducing Cekura Bench: the world’s leading Voice AI Benchmarks We benchmarked 50+ models from openAi, Google, Nvidia, xAI, Inworld, Gradium, Assembly, Fish audio and more Now you can: → Compare naturalness, performance and cost → See how the best voice agents are being built. → Find the best fit for what you are building. Our tests cover Speech-to-Speech models, Speech-to-Text models, Text-to-Speech Models, and Voice Agent platforms. We recorded real humans doing tasks in different domains and created the most representative voice benchmark out there! Explore the benchmarks: We would love to collaborate with more teams on these benchmarks. Reach out if you are interested.

Introducing Cekura Bench: the world’s leading Voice AI Benchmarks We benchmarked 50+ models from openAi, Google, Nvidia, xAI, Inworld, Gradium, Assembly, Fish audio and more Now you can: → Compare naturalness, performance and cost → See how the best voice agents are being built. → Find the best fit for what you are building. Our tests cover Speech-to-Speech models, Speech-to-Text models, Text-to-Speech Models, and Voice Agent platforms. We recorded real humans doing tasks in different domains and created the most representative voice benchmark out there! Explore the benchmarks: We would love to collaborate with more teams on these benchmarks. Reach out if you are interested.

Tarush Agarwal

798,909 views • 2 days ago

Realtime speech-to-text for developers building conversational apps, AI agents, and voice workflows.

Realtime speech-to-text for developers building conversational apps, AI agents, and voice workflows.

ElevenLabs Developers

15,174,464 views • 5 months ago

I've been building voice agents for the last 6mo and I think the chat-supervisor pattern is a game changer. Stitched model (STT-LLM-TTS) is slow, but realtime audio models aren't (yet) as smart as text. This has the best of both worlds. Here's how it works:

I've been building voice agents for the last 6mo and I think the chat-supervisor pattern is a game changer. Stitched model (STT-LLM-TTS) is slow, but realtime audio models aren't (yet) as smart as text. This has the best of both worlds. Here's how it works:

Noah MacCallum

73,682 views • 1 year ago

Voice agents get built three ways: cascaded (STT, LLM, TTS), turn-based speech-to-speech, and full-duplex, where the model listens and talks at the same time. All three now run on LiveKit Inference: GPT-Live and GPT Realtime from OpenAI, Grok Voice Think Fast from xAI, and Gemini Live from Google DeepMind. One LiveKit Cloud key, one bill.

Voice agents get built three ways: cascaded (STT, LLM, TTS), turn-based speech-to-speech, and full-duplex, where the model listens and talks at the same time. All three now run on LiveKit Inference: GPT-Live and GPT Realtime from OpenAI, Grok Voice Think Fast from xAI, and Gemini Live from Google DeepMind. One LiveKit Cloud key, one bill.

LiveKit

107,339 views • 23 days ago

Did xAI just mass-murder the entire voice AI industry? 🤯 Grok just launched two voice APIs. Speech-to-Text and Text-to-Speech. Built on the same stack powering Tesla cars and Starlink support. And priced at 10x cheaper than ElevenLabs. Speech-to-Text: $0.10/hr batch. $0.20/hr streaming. Text-to-Speech: $4.20 per million characters. 25+ languages. Real-time streaming. Speaker diarization. Already outperforming ElevenLabs, Deepgram, and AssemblyAI on word error rate. TTS ships with expressive tags like [laugh], [sigh], , . Voices that don't sound like robots reading a script. ElevenLabs spent years building a voice AI company. xAI built voice AI for cars and satellites.

Did xAI just mass-murder the entire voice AI industry? 🤯 Grok just launched two voice APIs. Speech-to-Text and Text-to-Speech. Built on the same stack powering Tesla cars and Starlink support. And priced at 10x cheaper than ElevenLabs. Speech-to-Text: $0.10/hr batch. $0.20/hr streaming. Text-to-Speech: $4.20 per million characters. 25+ languages. Real-time streaming. Speaker diarization. Already outperforming ElevenLabs, Deepgram, and AssemblyAI on word error rate. TTS ships with expressive tags like [laugh], [sigh], , . Voices that don't sound like robots reading a script. ElevenLabs spent years building a voice AI company. xAI built voice AI for cars and satellites.

Vaibhav Sisinty

24,539,570 views • 6 months ago

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

318,900 views • 1 year ago

How to build the world's fastest voice AI bot: - Self-host speech-to-text, LLM inference, and text-to-speech all together in the same container/cluster. - Route audio over the internet using WebRTC and edge networking. - Configure timings for voice activity detection, phrase endpointing, and other parts of the pipeline to optimize for latency. (There are trade-offs to doing this!) Here's a LLama 3 voice bot that has voice-to-voice response times of ~500ms. We used Deepgram's STT and TTS for this bot, and everything is hosted on cerebriumai's serverless GPU infrastructure.

How to build the world's fastest voice AI bot: - Self-host speech-to-text, LLM inference, and text-to-speech all together in the same container/cluster. - Route audio over the internet using WebRTC and edge networking. - Configure timings for voice activity detection, phrase endpointing, and other parts of the pipeline to optimize for latency. (There are trade-offs to doing this!) Here's a LLama 3 voice bot that has voice-to-voice response times of ~500ms. We used Deepgram's STT and TTS for this bot, and everything is hosted on cerebriumai's serverless GPU infrastructure.

kwindla

282,890 views • 2 years ago

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

329,439 views • 8 months ago

xAI just dropped 2 absolute monsters: Grok Speech-to-Text + Grok Text-to-Speech APIs Same battle-tested stack that runs Grok Voice, millions of Tesla cars, and Starlink support. - STT: Real-time WebSocket, speaker diarization, 25+ languages, word-level timestamps. - TTS: Super natural, expressive voices with tags for laugh, sigh, whisper, emphasis, pause; batch + streaming. Put voice AI on the list of things Grok already 100%'d xAI, Grok

xAI just dropped 2 absolute monsters: Grok Speech-to-Text + Grok Text-to-Speech APIs Same battle-tested stack that runs Grok Voice, millions of Tesla cars, and Starlink support. - STT: Real-time WebSocket, speaker diarization, 25+ languages, word-level timestamps. - TTS: Super natural, expressive voices with tags for laugh, sigh, whisper, emphasis, pause; batch + streaming. Put voice AI on the list of things Grok already 100%'d xAI, Grok

Mario Nawfal

164,897 views • 6 months ago

Eminem's fastest verse vs Smallest Pulse STT The last verse of "Godzilla" is one of the fastest in mainstream rap. Pulse kept up with all of it: 296 words, 1.4% word error rate. Just the raw audio, transcribed live. If Pulse can handle Eminem, your customer calls are easy.

Eminem's fastest verse vs Smallest Pulse STT The last verse of "Godzilla" is one of the fastest in mainstream rap. Pulse kept up with all of it: 296 words, 1.4% word error rate. Just the raw audio, transcribed live. If Pulse can handle Eminem, your customer calls are easy.

smallest.ai

28,570 views • 14 days ago

Introducing the Gemini API CLI, now live in preview! 🚀 The easiest way for your agents to generate images, videos, music, text-to-speech or use managed agents. A fast and efficient way to get access to all the goodness of the Gemini API. 🧵👇

Introducing the Gemini API CLI, now live in preview! 🚀 The easiest way for your agents to generate images, videos, music, text-to-speech or use managed agents. A fast and efficient way to get access to all the goodness of the Gemini API. 🧵👇

Jack Wotherspoon

22,901 views • 16 days ago