正在加载视频...

视频加载失败

返回首页

🎧 Audio book is now live on the site for MAKE book I ended up using the 30-second Instant Voice Clone by and just talked into my laptop a bit Instead of the Professional Voice Clone with the 30-minute cut from me on the Lex Fridman podcast, because it...

levelsio's profile picture

@levelsio

892,097 subscribers

173,286 次观看 • 1 年前 •via X (Twitter)

科学技术教育音乐
LIVE

Anya Rossi• Live Now

Private livecam show

11 条评论

AJ 的头像
AJ1 年前

@elevenlabsio personally I don't like it, it's not your vibe you sounds different for audio it's matters

@levelsio 的头像
@levelsio1 年前

@elevenlabsio Yes it's not there yet but a start

Brent Brookler 的头像
Brent Brookler1 年前

@elevenlabsio Did you know you can’t use an eleven labs audio for an Amazon or Apple audio book? I think it’s kinda crazy, but true story.

🧑🏻‍💻 Christopher 🎮 的头像
🧑🏻‍💻 Christopher 🎮1 年前

@elevenlabsio So, you didn't even read your own book....

Naxdee - e/acc 的头像
Naxdee - e/acc1 年前

@elevenlabsio The pauses (or lack of) always get me with Eleven Labs. I use them to make video content for my SaaS. If you find a way to fix this PLEASE let us know.

@levelsio 的头像
@levelsio1 年前

@elevenlabsio Yep pausing is a big issue with them @matistanis is aware and working on it I think

jack friks 的头像
jack friks1 年前

@elevenlabsio awesome you can do this since it’s your own platform hosting amazon/ audible is so picky and takes your rights (+ no use of ai voice allowed)

Rohit Mittal 的头像
Rohit Mittal1 年前

@elevenlabsio I tried ElevenLabs and translated the audio. It made us sound like Apu.

@levelsio 的头像
@levelsio1 年前

@elevenlabsio LMFAO

Kevin Naughton Jr. 的头像
Kevin Naughton Jr.1 年前

@elevenlabsio just a heads up that it doesn't seem clickable on mobile

@levelsio 的头像
@levelsio1 年前

@elevenlabsio Oops will fix thx

相关视频

So helped me to unlock the voice now it works! It's trained on my Lex Fridman podcast It's 99% perfect but the training data apparently still has some Lex "hummm" sounds in it that sometimes show up I think they call this "crosstalk" and it's hard to remove with AI so I think there's no other choice than to listen to the 1 hour of training data and remove those bits and re-train But then we have a good AI model to turn my book MAKE into an audio book without having to read it for hours/days/weeks
1:23

So helped me to unlock the voice now it works! It's trained on my Lex Fridman podcast It's 99% perfect but the training data apparently still has some Lex "hummm" sounds in it that sometimes show up I think they call this "crosstalk" and it's hard to remove with AI so I think there's no other choice than to listen to the 1 hour of training data and remove those bits and re-train But then we have a good AI model to turn my book MAKE into an audio book without having to read it for hours/days/weeks

@levelsio

169,787 次观看 • 1 年前

people always ask me how they can create long, consistent AI videos with audio. well creatify aurora solves a big part of that and there are some new models coming soon too for more action + audio. I created his voice with dia tts which i think great at capturing emotional speeches. then I cloned his voice using minimax voice clone and use it on minimax tts 2.6. here is the my full conversation with claude for images and text to speech inputs:
0:27

people always ask me how they can create long, consistent AI videos with audio. well creatify aurora solves a big part of that and there are some new models coming soon too for more action + audio. I created his voice with dia tts which i think great at capturing emotional speeches. then I cloned his voice using minimax voice clone and use it on minimax tts 2.6. here is the my full conversation with claude for images and text to speech inputs:

gokaygokay

38,341 次观看 • 8 个月前

Most voice models are great at cloning voices, but they still manage to erase the person. I have an accent. Voice models don't know how to preserve it. Whenever I clone my voice, models get the words right, but mess up the rhythm badly. They flatten my vowels, and everything drifts toward a generic American accent. I'm attaching one example from Mirage. This is the first model I have heard of that can reproduce access as-is. • My accent stays intact, even when speaking English and Spanish • My pacing and cadence feel intentional, not averaged • The expressivity of my voice comes through Best part: Mirage only needs a short sample of your voice to clone it. I think this solves the hard part of voice cloning: intelligibility is easy, but identity is what makes a huge difference. Listen to the attached sample, and you'll notice the difference if you compare it with any other model out there. If you want to try this out, check out this link: Thanks to the Mirage team for collaborating with me on this post.
0:21

Most voice models are great at cloning voices, but they still manage to erase the person. I have an accent. Voice models don't know how to preserve it. Whenever I clone my voice, models get the words right, but mess up the rhythm badly. They flatten my vowels, and everything drifts toward a generic American accent. I'm attaching one example from Mirage. This is the first model I have heard of that can reproduce access as-is. • My accent stays intact, even when speaking English and Spanish • My pacing and cadence feel intentional, not averaged • The expressivity of my voice comes through Best part: Mirage only needs a short sample of your voice to clone it. I think this solves the hard part of voice cloning: intelligibility is easy, but identity is what makes a huge difference. Listen to the attached sample, and you'll notice the difference if you compare it with any other model out there. If you want to try this out, check out this link: Thanks to the Mirage team for collaborating with me on this post.

Santiago

31,245 次观看 • 8 个月前

New trained AI clone of me here Nadav Keyson helped me again to improve the training data and remove the last parts of Lex's voice in there with Riverside.FM, very cool, now my voice's training data is clean! The voice sounds great I think, only problem I now have and I always have this with ElevenLabs already for years is that it somehow seems to NOT want to pause with periods and paragraphs It just keeps rambling like a steam train that can't be stopped How do I make my AI more chill? I don't wanna add a thousand times for every paragraph, seems like it should just be able to do this better no? @matistanis
0:44

New trained AI clone of me here Nadav Keyson helped me again to improve the training data and remove the last parts of Lex's voice in there with Riverside.FM, very cool, now my voice's training data is clean! The voice sounds great I think, only problem I now have and I always have this with ElevenLabs already for years is that it somehow seems to NOT want to pause with periods and paragraphs It just keeps rambling like a steam train that can't be stopped How do I make my AI more chill? I don't wanna add a thousand times for every paragraph, seems like it should just be able to do this better no? @matistanis

@levelsio

270,964 次观看 • 1 年前

🗣️: Your voice is so unique to me, the way you sing.. It’s very... it's very distinct. And sometimes it seems like I can hear the Western influences, but I also hear—and I think you told me—that there's a lot of Korean singers who inspired you as well. And sometimes it feels like you're less R&B influenced than the other members who sing and more just something very unique—a unique fusion. And I was wondering how you think about that fusion and style. 🐥: Well, my voice is just... I was just born with it. Beyond any influences, I just got it from my mother and father. But I'm so thankful for your kind words. Honestly, if anyone was an influence, I think the BTS members might have been my biggest influence because they've been hearing my voice all the way back since I was a trainee. They're the ones who told me I have a great voice and recommended songs for me to sing. Those recommendations have been very influential for me.
1:12

🗣️: Your voice is so unique to me, the way you sing.. It’s very... it's very distinct. And sometimes it seems like I can hear the Western influences, but I also hear—and I think you told me—that there's a lot of Korean singers who inspired you as well. And sometimes it feels like you're less R&B influenced than the other members who sing and more just something very unique—a unique fusion. And I was wondering how you think about that fusion and style. 🐥: Well, my voice is just... I was just born with it. Beyond any influences, I just got it from my mother and father. But I'm so thankful for your kind words. Honestly, if anyone was an influence, I think the BTS members might have been my biggest influence because they've been hearing my voice all the way back since I was a trainee. They're the ones who told me I have a great voice and recommended songs for me to sing. Those recommendations have been very influential for me.

cocojm⁷

14,792 次观看 • 3 个月前

noah is live on weverse with his hoarse voice rn😭 💜: my throat’s been pretty strained… 💜: i think i overused it because usually when i record, i try to capture all the little details 💜: so for the parts that require a lot of nuance, i have to use a lot of good vocal techniques. because of that my voice is really worn out right now. 💜: it sounds pretty hoarse, right? 💜: but well… u guys like it when my voice is hoarse, don’t u? 💜: it sounds sexy u said. ha ha ha. HE FREAKING KNEW.
0:41

noah is live on weverse with his hoarse voice rn😭 💜: my throat’s been pretty strained… 💜: i think i overused it because usually when i record, i try to capture all the little details 💜: so for the parts that require a lot of nuance, i have to use a lot of good vocal techniques. because of that my voice is really worn out right now. 💜: it sounds pretty hoarse, right? 💜: but well… u guys like it when my voice is hoarse, don’t u? 💜: it sounds sexy u said. ha ha ha. HE FREAKING KNEW.

데이🌻

11,438 次观看 • 16 天前

. T I don’t know if you’ll ever see this, but I spent almost two weeks making a just my voice version of CHROMAKOPIA and it just got spontaneously blocked this morning. In this video, it is CHROMAKOPIA but every single sound is made using my voice, including the drums and bass even. I’ve done these videos for a while and sometimes they go viral. If there is any way you can get this unblocked it would mean the world to me, I spent over 150 hours making this and it's the best piece of music I've ever been a part of and would hate for no one to be able to watch it. Below I’ve attached Thought I Was Dead but it’s just my voice as an example for what the content is.
1:46

. T I don’t know if you’ll ever see this, but I spent almost two weeks making a just my voice version of CHROMAKOPIA and it just got spontaneously blocked this morning. In this video, it is CHROMAKOPIA but every single sound is made using my voice, including the drums and bass even. I’ve done these videos for a while and sometimes they go viral. If there is any way you can get this unblocked it would mean the world to me, I spent over 150 hours making this and it's the best piece of music I've ever been a part of and would hate for no one to be able to watch it. Below I’ve attached Thought I Was Dead but it’s just my voice as an example for what the content is.

Spectre

1,148,260 次观看 • 1 年前

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.
1:33

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 次观看 • 4 个月前

2min love the sound of each other's voices and seungmin wrote cinema with a specific sound of lee know's voice in mind 😭😭😭 🐶: to add a bit about "cinema", there's this voice of yours that i really like and that voice stay really loves it too. that old, kind of tickly natural sound you just let out. i wanted to bring that back after a long time. so i wrote a lot while thinking about that. even when asked 3racha for advice, i asked a lot about which direction to pull it 🐶: so it would feel like "does this song and you blend well?" after i finished the song, recorded everything and played it for 3racha they said "did you write this thinking about lee know?" 🐰: back then, when i first heard seungmin's voice, early on, i remember thinking, "nice tone". the way he does it calmly like that, i always really liked listening to it
0:46

2min love the sound of each other's voices and seungmin wrote cinema with a specific sound of lee know's voice in mind 😭😭😭 🐶: to add a bit about "cinema", there's this voice of yours that i really like and that voice stay really loves it too. that old, kind of tickly natural sound you just let out. i wanted to bring that back after a long time. so i wrote a lot while thinking about that. even when asked 3racha for advice, i asked a lot about which direction to pull it 🐶: so it would feel like "does this song and you blend well?" after i finished the song, recorded everything and played it for 3racha they said "did you write this thinking about lee know?" 🐰: back then, when i first heard seungmin's voice, early on, i remember thinking, "nice tone". the way he does it calmly like that, i always really liked listening to it

lena ✰

14,277 次观看 • 7 个月前

My first book publishes today. It has nothing to do with hardcore finance or bitcoin, but everything to do with self-improvement, constant learning, and living the life you want. The book, which is titled How To Live An Extraordinary Life, is a compilation of 65 letters to my children on various life lessons I have picked up over the years. I never had an intention to write a book, but as I started writing these letters in private it struck me that many other people could benefit from these insights. The beauty of the book is that each letter is the product of a mistake I made, a piece of advice someone gave to me, or an experience I lived through over the years. You will learn lessons about money, investing, work, relationships, health, and happiness. None of the ideas are my own — I had to live my life and rely on people much smarter than me to teach me all of this information. Each letter in the book is structured to share the lesson, explain how I learned it, and then give actionable advice on how to implement it in your own life. My goal is to share the information with my children and some of you. These are timeless lessons that have been passed down to me, so their survival over time signals their accuracy and value. As you all know, I hate asking people for a favor. But the publisher has told me that sales in the first week really matter for the long-term popularity of the book. With that understanding, if I have ever helped you in any way (taught you something, given you an idea, made you think more critically, etc), then please consider buying the book today. BUY NOW:
1:19

My first book publishes today. It has nothing to do with hardcore finance or bitcoin, but everything to do with self-improvement, constant learning, and living the life you want. The book, which is titled How To Live An Extraordinary Life, is a compilation of 65 letters to my children on various life lessons I have picked up over the years. I never had an intention to write a book, but as I started writing these letters in private it struck me that many other people could benefit from these insights. The beauty of the book is that each letter is the product of a mistake I made, a piece of advice someone gave to me, or an experience I lived through over the years. You will learn lessons about money, investing, work, relationships, health, and happiness. None of the ideas are my own — I had to live my life and rely on people much smarter than me to teach me all of this information. Each letter in the book is structured to share the lesson, explain how I learned it, and then give actionable advice on how to implement it in your own life. My goal is to share the information with my children and some of you. These are timeless lessons that have been passed down to me, so their survival over time signals their accuracy and value. As you all know, I hate asking people for a favor. But the publisher has told me that sales in the first week really matter for the long-term popularity of the book. With that understanding, if I have ever helped you in any way (taught you something, given you an idea, made you think more critically, etc), then please consider buying the book today. BUY NOW:

Anthony Pompliano 🌪

1,369,171 次观看 • 1 年前

#BaiLu on choosing to dub Li Peiyi with her own voice “I’d like to talk a bit about the dubbing. It was mainly because of the filming environment in Hengdian there were a lot of noise issues. When we first started shooting this drama, the plan was to use the original on-set audio entirely. But as filming continued, the environmental interference became too much. We had a huge amount of dialogue every day, and eventually the director told us, ‘Forget it either you dub it yourselves in post-production, or we’ll bring in professional voice actors.’ By the end, voice actress Qiao Shiyu had already recorded my lines three times. She worked incredibly hard! Later on, though, the director messaged me on WeChat and said, ‘Lulu, come to the editing room. Don’t just listen on your phone listen in surround sound and then make your choice.’ Since this character is somewhat similar to my role in The Legends (Zhao Yao), Teacher Qiao Shiyu had dubbed it very diligently three times. In the end, I spent an entire afternoon in the editing room, and the whole team decided together that my natural voice might suit Pei Yi better. My voice isn’t perfect, and my delivery isn’t the absolute best, but perhaps for Pei Yi who isn’t meant to sound overly polished or perfectly ‘pretty’ my vocal tone fit the character more. So we ultimately chose to use my own voice. As actors, we know that using original audio is a real test of our abilities. We’ll keep working hard, and for any areas where I didn’t do well enough, I hope everyone can be understanding.”
1:36

#BaiLu on choosing to dub Li Peiyi with her own voice “I’d like to talk a bit about the dubbing. It was mainly because of the filming environment in Hengdian there were a lot of noise issues. When we first started shooting this drama, the plan was to use the original on-set audio entirely. But as filming continued, the environmental interference became too much. We had a huge amount of dialogue every day, and eventually the director told us, ‘Forget it either you dub it yourselves in post-production, or we’ll bring in professional voice actors.’ By the end, voice actress Qiao Shiyu had already recorded my lines three times. She worked incredibly hard! Later on, though, the director messaged me on WeChat and said, ‘Lulu, come to the editing room. Don’t just listen on your phone listen in surround sound and then make your choice.’ Since this character is somewhat similar to my role in The Legends (Zhao Yao), Teacher Qiao Shiyu had dubbed it very diligently three times. In the end, I spent an entire afternoon in the editing room, and the whole team decided together that my natural voice might suit Pei Yi better. My voice isn’t perfect, and my delivery isn’t the absolute best, but perhaps for Pei Yi who isn’t meant to sound overly polished or perfectly ‘pretty’ my vocal tone fit the character more. So we ultimately chose to use my own voice. As actors, we know that using original audio is a real test of our abilities. We’ll keep working hard, and for any areas where I didn’t do well enough, I hope everyone can be understanding.”

美

21,805 次观看 • 6 个月前

There's a very small YouTube channel that uses stock footage and narration, where the voice seems to be AI trained on my voice. I was told about it from one of my viewers that said when he started watching this channel he realized it was my voice, but with AI. It for sure doesn't sound perfect, and it still has that "AI" sound to it, but this stuff is going to get a lot more convincing really soon. There's not really any tools to deal with this currently, as there's no reporting system, and it's not an exact copy because the narration is different. But I feel like this may become a much bigger problem for YouTube in the future. They are going to have to adapt their copyright system to include AI voice and video copies of people under impersonation.
0:37

There's a very small YouTube channel that uses stock footage and narration, where the voice seems to be AI trained on my voice. I was told about it from one of my viewers that said when he started watching this channel he realized it was my voice, but with AI. It for sure doesn't sound perfect, and it still has that "AI" sound to it, but this stuff is going to get a lot more convincing really soon. There's not really any tools to deal with this currently, as there's no reporting system, and it's not an exact copy because the narration is different. But I feel like this may become a much bigger problem for YouTube in the future. They are going to have to adapt their copyright system to include AI voice and video copies of people under impersonation.

Joseph Carlson

46,024 次观看 • 1 年前

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)
0:40

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)

kwindla

40,593 次观看 • 2 个月前

Marc Andreessen: "I'm reading every spare minute that I have" Expanding on his information diet, the a16z cofounder says AirPods have been the single biggest technological leap: "They're the unlock for me for audiobooks, podcasts, and interviews. I'm doing audio content probably 2-3 hours per day -- getting up in the morning, going to bed at night, all the drive time . . . If nothing else is going on, I'm always listening to something." Text-to-speech has also been an unlock with apps like Substack, NaturalReader, and Apple's audiobooks -- "It sounds spectacular," Marc says, "[AI voice technology] is really starting to work." Marc uses podcasts and YouTube interviews to go down rabbit holes on topics, but otherwise he's trying to get back to audiobooks: "I try to get back to audiobooks as much as possible, and the reason is audiobooks are my opportunity to really learn a new area that I probably don't know anything about. And so if I can scrape aside 10-20 hours of audio time for a period of history or something like that, I can really go deep on it." Marc generally tries to barbell his information intake: "It's either stuff that's super current or it's stuff that's timeless. I'm basically trying to not read anything that's from yesterday through like 10 years ago. I'm trying to be super current, and the form of being super current is talking to people who are currently experts or it's Twitter. Then for timeless, that's almost all books, but I kind of go back and forth between these modes." He continues: "I'm either listening to a book that's usually on history or a biography or something like that or some new domain that I'm trying to learn. Or I'm up to the minute on what's happening in AI today." source: David Perell (2023)
2:49

Marc Andreessen: "I'm reading every spare minute that I have" Expanding on his information diet, the a16z cofounder says AirPods have been the single biggest technological leap: "They're the unlock for me for audiobooks, podcasts, and interviews. I'm doing audio content probably 2-3 hours per day -- getting up in the morning, going to bed at night, all the drive time . . . If nothing else is going on, I'm always listening to something." Text-to-speech has also been an unlock with apps like Substack, NaturalReader, and Apple's audiobooks -- "It sounds spectacular," Marc says, "[AI voice technology] is really starting to work." Marc uses podcasts and YouTube interviews to go down rabbit holes on topics, but otherwise he's trying to get back to audiobooks: "I try to get back to audiobooks as much as possible, and the reason is audiobooks are my opportunity to really learn a new area that I probably don't know anything about. And so if I can scrape aside 10-20 hours of audio time for a period of history or something like that, I can really go deep on it." Marc generally tries to barbell his information intake: "It's either stuff that's super current or it's stuff that's timeless. I'm basically trying to not read anything that's from yesterday through like 10 years ago. I'm trying to be super current, and the form of being super current is talking to people who are currently experts or it's Twitter. Then for timeless, that's almost all books, but I kind of go back and forth between these modes." He continues: "I'm either listening to a book that's usually on history or a biography or something like that or some new domain that I'm trying to learn. Or I'm up to the minute on what's happening in AI today." source: David Perell (2023)

Perch

341,799 次观看 • 7 个月前

Yejun 280726 Weverse Live 💙: I ended up coming on Weverse by myself like this 💙: I was in the middle of recording but I finished my part first completely and came here right away 💙: The other members are still in the studio, they're either recording, waiting for their turn, or practicing 💙: I can't really tell you to look forward to it too much because we're still working on it but... I wanted to tell you that we're in the middle of preparing it 💙: My voice is a little hoarse right now isn't it? It's okay though, honestly it's just that my throat is a bit swollen, it'll go down after about 30 minutes and I'll be fine 💙: But I thought it might be a little uncomfortable for PLLIs to listen to so please understand
0:46

Yejun 280726 Weverse Live 💙: I ended up coming on Weverse by myself like this 💙: I was in the middle of recording but I finished my part first completely and came here right away 💙: The other members are still in the studio, they're either recording, waiting for their turn, or practicing 💙: I can't really tell you to look forward to it too much because we're still working on it but... I wanted to tell you that we're in the middle of preparing it 💙: My voice is a little hoarse right now isn't it? It's okay though, honestly it's just that my throat is a bit swollen, it'll go down after about 30 minutes and I'll be fine 💙: But I thought it might be a little uncomfortable for PLLIs to listen to so please understand

nanas

16,008 次观看 • 17 天前

Today we’ve raised $52M Seed and we are announcing the public launch of S2.1 Pro. >It can clone a voice from 5 seconds of audio >2x faster than Cartesia & 1/6th the cost of Eleven Labs >most expressive model with word level control over emotion, intonation, pacing etc We support frontier AI companies including HeyGen, LiveKit, Retell, Sanas, and OpenArt all run our model in production. If you're a business and we can't cut your voice AI costs by 50%, we'll give you 1 year of Fish Audio for free. Book a demo: To celebrate our first birthday, we'll give you 1 month of S2.1 Pro for free. Like, retweet, and comment “Fish” to get it.
2:23

Today we’ve raised $52M Seed and we are announcing the public launch of S2.1 Pro. >It can clone a voice from 5 seconds of audio >2x faster than Cartesia & 1/6th the cost of Eleven Labs >most expressive model with word level control over emotion, intonation, pacing etc We support frontier AI companies including HeyGen, LiveKit, Retell, Sanas, and OpenArt all run our model in production. If you're a business and we can't cut your voice AI costs by 50%, we'll give you 1 year of Fish Audio for free. Book a demo: To celebrate our first birthday, we'll give you 1 month of S2.1 Pro for free. Like, retweet, and comment “Fish” to get it.

Fish Audio

6,179,447 次观看 • 16 天前

someone is going to make millions with this in 2026 99% of people think this is a real human (or they fail to notice it’s an ai-generated video) but this video is completely ai-generated, including the background music. let me teach you how to create this in a few minutes follow this workflow step by step: first, create the base image of your ai influencer using nb pro. this is currently the best tool for character consistency i used a json prompt to generate the base image then i turn it into video (i will share the exact json prompt with you in this thread) now, to generate the video paste this prompt into google veo using the “frame to video” option “a man in his 40s sits on a 1980s living room couch, looking directly at the camera with a serious expression. he gestures naturally with one hand as he speaks in vintage tv broadcast aesthetic: "today is october 12th, 1985. what i'm about to tell you will sound impossible... but mark my words, these three predictions will come true. no background music, no sound effects” next, i gave it a chunk of script, i only changed the script dialogue each time and kept the rest of the prompt exactly the same using this method, you also get a little consintent voiceover here is the format for you to use: [character description] + [visual style] + [dialogue of your script for under 8 seconds] for example, for the next part dialogue; [A man in his 40s sits on a 1980s living room couch, looking directly at the camera with a serious expression. He gestures naturally with one hand as he speaks in Vintage TV broadcast aesthetic: "but mark my words, these three predictions will come true... ONE: You will carry a device no bigger than a playing card that holds ten thousand songs" ] rest things i adjusted in the editing, now how to clone the audio for these several clips we just generated? i got the best audio from the very first clip i generated using veo 3 but here is the trick: - export that clip to a video editor - detach the audio of it - duplicate it to make it 10+ seconds long - clone the voice over using elevenlabs (go to 11Labs-> click on voices-> click on "create or clone a voice" button in the top right side) (i named mine “1985 ai influencer” inside 11labs) then, finally export all your video clips into your video editor detach the audios of all clips, and export it to 11Labs to clone it with that "1985 AI Influencer" voice once you dubbed it, import it back to your editor that’s it. there are endless use cases where you can use such an ai influencer like this to promote your biz: - skincare - weight loss & nutrition - psychology and mental health - marketing and sales - predictions (like this video) - making money & career growth - dating, parenting, and so on… there are a few people already started using such ai influencers, you are just behind them, you can find their pages on instagram don't be lazy, create one such an ai influencer for your targeted biz 2026 is going to be yours
0:40

someone is going to make millions with this in 2026 99% of people think this is a real human (or they fail to notice it’s an ai-generated video) but this video is completely ai-generated, including the background music. let me teach you how to create this in a few minutes follow this workflow step by step: first, create the base image of your ai influencer using nb pro. this is currently the best tool for character consistency i used a json prompt to generate the base image then i turn it into video (i will share the exact json prompt with you in this thread) now, to generate the video paste this prompt into google veo using the “frame to video” option “a man in his 40s sits on a 1980s living room couch, looking directly at the camera with a serious expression. he gestures naturally with one hand as he speaks in vintage tv broadcast aesthetic: "today is october 12th, 1985. what i'm about to tell you will sound impossible... but mark my words, these three predictions will come true. no background music, no sound effects” next, i gave it a chunk of script, i only changed the script dialogue each time and kept the rest of the prompt exactly the same using this method, you also get a little consintent voiceover here is the format for you to use: [character description] + [visual style] + [dialogue of your script for under 8 seconds] for example, for the next part dialogue; [A man in his 40s sits on a 1980s living room couch, looking directly at the camera with a serious expression. He gestures naturally with one hand as he speaks in Vintage TV broadcast aesthetic: "but mark my words, these three predictions will come true... ONE: You will carry a device no bigger than a playing card that holds ten thousand songs" ] rest things i adjusted in the editing, now how to clone the audio for these several clips we just generated? i got the best audio from the very first clip i generated using veo 3 but here is the trick: - export that clip to a video editor - detach the audio of it - duplicate it to make it 10+ seconds long - clone the voice over using elevenlabs (go to 11Labs-> click on voices-> click on "create or clone a voice" button in the top right side) (i named mine “1985 ai influencer” inside 11labs) then, finally export all your video clips into your video editor detach the audios of all clips, and export it to 11Labs to clone it with that "1985 AI Influencer" voice once you dubbed it, import it back to your editor that’s it. there are endless use cases where you can use such an ai influencer like this to promote your biz: - skincare - weight loss & nutrition - psychology and mental health - marketing and sales - predictions (like this video) - making money & career growth - dating, parenting, and so on… there are a few people already started using such ai influencers, you are just behind them, you can find their pages on instagram don't be lazy, create one such an ai influencer for your targeted biz 2026 is going to be yours

ViralOps

14,320 次观看 • 8 个月前

✨ A dream I had finally came true: I can now chat directly with my sites to build any feature or fix any bug just via Telegram I've been playing with OpenClaw for 3 weeks now and it's great but I was always too scared to run it on any production server And I was right a bit as Marc Köhlbrugge was able to hack it by social engineering and acting as if it was me, and with enough tries it believed him, and was able to modify the server, change SSH keys etc. of course I had it isolated properly on its own VPS and it didn't touch anything sensitive (as it should!) Marc then reported that bug to Peter Steinberger 🦞 who patched it fast But I wanted to try something more basic and simple, and I think maybe more secure: to just connect Claude Code on my server to Telegram which would be hard locked to only messages from me So I installed claude-code-telegram by Richard Atkinson on the server and run it as a system daemon and it works really well The cool thing is that I was already using Telegram for server errors like this: > Photo AI - ❌ Random credits giveaway failed (Attempt 30/30) with an exception: SQLSTATE[HY000]: General error: 5 database is locked So now I can just reply, "Ok fix this", and Claude Code on the server in production will try (and probably succeed) in fixing it In the video below I asked it to make show [🌳 Parks ] on the map by default on load, it did that, then I reloaded the page and it instantly worked One thing it still needs is sending actual messages while it's doing stuff which OpenClaw does really well, it's annoying to just wait while it says "Working..." but that's probably next
1:30

✨ A dream I had finally came true: I can now chat directly with my sites to build any feature or fix any bug just via Telegram I've been playing with OpenClaw for 3 weeks now and it's great but I was always too scared to run it on any production server And I was right a bit as Marc Köhlbrugge was able to hack it by social engineering and acting as if it was me, and with enough tries it believed him, and was able to modify the server, change SSH keys etc. of course I had it isolated properly on its own VPS and it didn't touch anything sensitive (as it should!) Marc then reported that bug to Peter Steinberger 🦞 who patched it fast But I wanted to try something more basic and simple, and I think maybe more secure: to just connect Claude Code on my server to Telegram which would be hard locked to only messages from me So I installed claude-code-telegram by Richard Atkinson on the server and run it as a system daemon and it works really well The cool thing is that I was already using Telegram for server errors like this: > Photo AI - ❌ Random credits giveaway failed (Attempt 30/30) with an exception: SQLSTATE[HY000]: General error: 5 database is locked So now I can just reply, "Ok fix this", and Claude Code on the server in production will try (and probably succeed) in fixing it In the video below I asked it to make show [🌳 Parks ] on the map by default on load, it did that, then I reloaded the page and it instantly worked One thing it still needs is sending actual messages while it's doing stuff which OpenClaw does really well, it's annoying to just wait while it says "Working..." but that's probably next

@levelsio

642,596 次观看 • 5 个月前