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。

Today, we're ending boring transcripts. Fish Audio’s ASR model now: → Identifies different speakers → Understands what they're feeling → Labels emotion cues like [laughter], [surprised] in the transcript All this while supporting 83 languages and being the most accurate STT model out there. Try it out now 🐟

Fish Audio

35,757 subscribers

44,945 次观看 • 1 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

19 条评论

OIiver 的头像

OIiver1 个月前

Better than Meta muse transcribe?

Fish Audio 的头像

Fish Audio1 个月前

feel free to try out!

OIiver 的头像

OIiver1 个月前

Ah it’s looking like it’s not streaming :(

Resona 的头像

Resona1 个月前

@soniox_ai I am certain your accuracy, especially for non English languages is far superior. However, this is a novel concept for adding significantly more context to transcribed speech. Hopefully we get something like this in Soniox v6.

Sam 的头像

Sam1 个月前

any plan to open source?

Vishal Singh 🥑 的头像

Vishal Singh 🥑1 个月前

Excited to try this ✨

Mohammed 的头像

Mohammed1 个月前

I smell competition @ElevenLabs

XXXTENTACIOFF 的头像

XXXTENTACIOFF1 个月前

Not even close to @soniox_ai lol

Marsonal 的头像

Marsonal1 个月前

is it free?

Adam Gera 亚当的头像

Adam Gera 亚当1 个月前

Make it cheaper.

JK | O ULTIMO PARDO | EDITOR 的头像

JK | O ULTIMO PARDO | EDITOR1 个月前

BY OPPUS 5.5

Chase Chick 的头像

Chase Chick1 个月前

Awesome! Is this going to be integrated into the agents?

Kaylard - e/acc 的头像

Kaylard - e/acc1 个月前

We need benchmarks

Flix Muller 的头像

Flix Muller1 个月前

This is a big upgrade for transcripts. Speaker detection + emotion cues + 83 languages makes ASR much more useful for real conversations. 🔥

Daniel Priscu 的头像

Daniel Priscu1 个月前

emotion labels in the transcript are going to make meeting notes a lot more awkward

JimRat 的头像

JimRat1 个月前

Whats the open source plans?

Alex 的头像

Alex1 个月前

Please add more languages and accuracy

Emmanuel 的头像

Emmanuel1 个月前

@grok what language do they support? Any aftican language

LastResort 的头像

LastResort1 个月前

Price? @grok

相关视频

The Devin CLI has a brand new model picker! You can now set your model, tune its reasoning effort, and compare costs - all in one view Try it today with `/model`

Devin Developers

28,495 次观看 • 2 个月前

Drama 3 by Fish Audio is now the top open TTS model. Allows meta tags within your transcript to control emotion, direction, SFX. Pretty good at voice cloning too.

Drama 3 by Fish Audio is now the top open TTS model. Allows meta tags within your transcript to control emotion, direction, SFX. Pretty good at voice cloning too.

AI Search

22,923 次观看 • 27 天前

Not only did Trixie and Katya watch Reality Check: Inside America's Next Top Model... they're in it! A new episode of I Like To Watch is out now

Not only did Trixie and Katya watch Reality Check: Inside America's Next Top Model... they're in it! A new episode of I Like To Watch is out now

Most

185,464 次观看 • 8 个月前

1/ First, how do you benchmark an AI model? You give different models the same test cases, and score them. You can do the same with voice models using Word Error Rate (WER). You generate the audio, transcribe it with a speech recognition (ASR) model, and compare that transcript to the original text. And this tells you whether the model said the right words. But the same sentence can be said correctly, while still being wrong in context.

1/ First, how do you benchmark an AI model? You give different models the same test cases, and score them. You can do the same with voice models using Word Error Rate (WER). You generate the audio, transcribe it with a speech recognition (ASR) model, and compare that transcript to the original text. And this tells you whether the model said the right words. But the same sentence can be said correctly, while still being wrong in context.

Cartesia

10,384 次观看 • 1 个月前

We just killed Pangram, the world's most accurate AI detector. Today we’re announcing Super, a next generation humanization model. Our model makes content so unique and natural AI checkers can't even detect it. Check this out ⬇️

We just killed Pangram, the world's most accurate AI detector. Today we’re announcing Super, a next generation humanization model. Our model makes content so unique and natural AI checkers can't even detect it. Check this out ⬇️

StealthGPT

448,190 次观看 • 2 个月前

Gradium TTS beta model is out now to define what fast means. Less than 50ms TTFA, same quality. Most models are >100ms TTFA. Try it now:

Gradium TTS beta model is out now to define what fast means. Less than 50ms TTFA, same quality. Most models are >100ms TTFA. Try it now:

Gradium

24,860 次观看 • 1 个月前

Fishing? We're kinda obsessed, and this is just the beginning. Quick question: What type of fish would you like to see in the game? 🐟🐠🦈 Pre-register NOW: #DestinyRising #PreRegisterNow

Fishing? We're kinda obsessed, and this is just the beginning. Quick question: What type of fish would you like to see in the game? 🐟🐠🦈 Pre-register NOW: #DestinyRising #PreRegisterNow

Destiny: Rising

262,902 次观看 • 1 年前

Nano Banana is finally here. the most powerful image editing AI model is now available in Krea. try it now!

Nano Banana is finally here. the most powerful image editing AI model is now available in Krea. try it now!

Krea

109,844 次观看 • 1 年前

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

123,084 次观看 • 3 个月前

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

239,584 次观看 • 2 个月前

Cohere's Model Vault is now available in public beta. With NVIDIA AI Confidential Computing, and GPUs on Microsoft Azure, anybody can run our models while keeping their information secure. Try it out today:

Cohere's Model Vault is now available in public beta. With NVIDIA AI Confidential Computing, and GPUs on Microsoft Azure, anybody can run our models while keeping their information secure. Try it out today:

Cohere

16,812 次观看 • 19 天前

We pioneered the first ultra-realistic Text to Speech model, and recently launched the world's most accurate Speech to Text model, Scribe. But we're not stopping there. Today, we're taking one small step for man, and one giant leap for man's best friend... with Text to Bark.

We pioneered the first ultra-realistic Text to Speech model, and recently launched the world's most accurate Speech to Text model, Scribe. But we're not stopping there. Today, we're taking one small step for man, and one giant leap for man's best friend... with Text to Bark.

ElevenLabs

291,881 次观看 • 1 年前

Today we're releasing Claude 3 Haiku, the fastest and most affordable model in its intelligence class. Haiku is now available in the API and on for Claude Pro subscribers.

Today we're releasing Claude 3 Haiku, the fastest and most affordable model in its intelligence class. Haiku is now available in the API and on for Claude Pro subscribers.

Anthropic

465,705 次观看 • 2 年前

Introducing Eleven v3 (alpha) - the most expressive Text to Speech model ever. Supporting 70+ languages, multi-speaker dialogue, and audio tags such as [excited], [sighs], [laughing], and [whispers]. Now in public alpha and 80% off in June.

Introducing Eleven v3 (alpha) - the most expressive Text to Speech model ever. Supporting 70+ languages, multi-speaker dialogue, and audio tags such as [excited], [sighs], [laughing], and [whispers]. Now in public alpha and 80% off in June.

ElevenLabs

1,960,325 次观看 • 1 年前

Today we're rolling out Dreamina Seedance 2.5 in the U.S., and it's now available worldwide on CapCut. The full model: 30 seconds from a single generation, up to 50 references, and precise edits to any part of a shot. Available across App, Web, and Desktop. What will you create? Try it now. RT+Comment in 4hr to get extra 1000 credit in your DM!

Today we're rolling out Dreamina Seedance 2.5 in the U.S., and it's now available worldwide on CapCut. The full model: 30 seconds from a single generation, up to 50 references, and precise edits to any part of a shot. Available across App, Web, and Desktop. What will you create? Try it now. RT+Comment in 4hr to get extra 1000 credit in your DM!

CapCut

2,753,060 次观看 • 3 个月前

Today we’re launching Voice 2.0 in Chatbase. Voice agents used to make you choose: - Sound human with a voice-to-voice model - Or be smart with STT → LLM → TTS Now you get both. A voice model listens and talks, so it sounds like a real person and replies fast, and delegates the thinking to a frontier model like Opus 5.5, so it follows your Procedures, calls Actions reliably and stays inside your Guardrails. Try it at

Today we’re launching Voice 2.0 in Chatbase. Voice agents used to make you choose: - Sound human with a voice-to-voice model - Or be smart with STT → LLM → TTS Now you get both. A voice model listens and talks, so it sounds like a real person and replies fast, and delegates the thinking to a frontier model like Opus 5.5, so it follows your Procedures, calls Actions reliably and stays inside your Guardrails. Try it at

Yasser

37,292 次观看 • 9 天前

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 👇

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

102,194 次观看 • 15 天前