This is perhaps the most useful tutorial on AI video generation I have watched the past 6 months. Thank you Jonathan. The world needs people like you. Keep it up. I really need this information. Subscribed and liked.
Wow, that is quite a compliment! That really is great to hear. Very motivating indeed. I always try to do my best. For some reason this video has done better than all my previous ones, but I always do things to this level. So glad to hear it was useful to you. Soon I'll also be doing a video on making the podcast with your own avatar in it, as well as a new tool I found that allows you to change the voice etc. Best regards and thanks for watching, commenting, and subscribing!
@@Jonathans-HubJam Brilliant work, but I wonder, why there is not a single tool in this AI era, that can separate male pitch from a female pitch? I think this this is easy but no body bother to create the one such,
@@librakhan25 Yes, surely those that know about ai voice technology could do that easily! I hope they do it soon.. Glad you liked the video by the way. Best regards, Jonathan.
Been doing the audio bit for a while now and have a channel on Spotify and use this also for my work. With a little creativity, one can take this quite a long way. And I guess, this is only the beginning. Thanks for enhancing this with the video side of it.
Hi Jonathan, I've been experimenting with it for two weeks now and yes, it's been well received in my LEGO niche and my local language. Unfortunately, the tool still has extreme accents that are difficult to get rid of in German. But it's fascinating beyond belief.
Thanks for watching and commenting! Oh very interesting, LEGO, eh? So you're making videos from the Notebook dialogue? And are you using HeyGen for the voices? Regards.
@@Jonathans-HubJam Hello Jonathan, no, I use other people's videos that have information about LEGO products that we don't have here in Germany yet. That means I take the transcription, translate it into German and paste it into Notebook and have a dialogue/podcast produced. It usually works the first time. The soundtrack is then edited and cut accordingly. Then it is published.
Oh I see, that's a use case I'd never have thought of! I haven't tried it in other languages yet. I speak fluent Spanish so want to give that a try. By the way I am very soon presenting a new couple of tools you might be interested in - one that makes a complete two person talking video directly from a podcast audio,, and another which is an ai podcast creation tool that gives you complete customisation and control over voices and content. Look out for it! Best regards.
@@Jonathans-HubJam Yes, here I am once again way ahead in the LEGO niche. Some people in the bubble are trying to use Suno to create background music or chatgtp teasers. But I claim to be putting in more effort and experimenting here. Now, I'm excited to see you produce a complete AI-generated video with the audio podcast! You have my subscription anyway. I would of course be delighted to get the famous click back from you. Maybe you can even use one of the podcast videos as an example to show that others are already trying it. Because I and my small team are already working on making the image avatars speak.
@@GenialesTuning Subscribed! Also my next video (talking about my viewer's suggestions for improvements to the process) will introduce this service of making a video from a notebook podcast (It's not available for people to use yet, but should be any day now), then I'll do a more in-depth video on it.
I've used Hedra a few times to try and make a singing avatar for my Suno songs, using their provided "isolated vocals" (which aren't the cleanest, but are decent). It's never worked out very well, with the movements and lip sync being pretty low-quality. From your video, it looks like it handles clean speech much better than singing, though. It's an impressive result, for sure.
Chopping up the audio to separate 2 speakers sounds like a labour intensive task. Perhaps using Speaker Diarization tools, for example, from Assembly AI would help speed things up.
Hi, talkingavatar has a Podcast Avatar feature, which could automatically diarizing the speaker, matching audio segments to the speaker, and ensuring lip-sync Podcast Video. With this new feature, you can generate a Podcast Video directly from NotebookLM audio with just one click, which is much more convenient.
Thank you Jonathan for this wonderful class! I loved it! I wonder if there is a way to generate the same video with different languages or generate the subtitles in foreign languages so you can reach much more audience. Thank you again. At this very moment my choice would be the Hedra mode. 🙏
Thanks for watching and thanks for the compliment. So glad to hear you loved it! You could transcribe the audio from Notebook by using Descript, then use the transcription to generate speech with a text-to-speech-service, such as Eleven Labs, then you'd make the avatar video as in my video. Alternatively you could just use Eleven Labs "Dubbing Studio" which takes any video or audio and allows you to dub it into a different language, so you could either make the original language version avatar video first, then dub it, or you could start with the audio from NotebookLM and import it into the Dubbing Studio to make other videos from that in other languages. I think Eleven Labs' dubbing studio is still quite unknown but it's really powerful, check it out: bit.ly/HubJam-ElevenLabs
@@Jonathans-HubJam So kind of you answering me. Thank you for your response. At this very moment I am at my last PhD year, and my thesis is about AI in Education and Teacher Education. So, I am learning and intend to work with that, helping others to improve their teaching and learning processes. Thank you again! A big Brazilian hug! 🙏
Yeah it is a bit mind-numbing, but not too bad once you get into the flow. The trick is to try to get Notebook to output not too long - ie 20mins would take much longer to edit than 10, so try to keep the source as concise as possible. Regards.
@Jonathans-HubJam agree! However, I'm using NLM for doing book reviews, mostly about Food, Cooking and the Science behind it. So the podcasts will be on average, 15 - 20 minutes. 🙂 However, I'm ready to put in the grind 😁.
If you're interested in a little bit more creative control in the editing/creation process, would love for you to check us out! NotebookLM for Creators
Thanks for the video. Is it possible to have the voice saved and the avatar (host of podcast) saved too? To create consistent audio and video with the same Avatar (host of podcast) but different guests?
I was sure I'd answered this, but looks like I didn't send.. if I understand your question right, the avatars in Heygen are always available in their library, and the Hedra avatars are available in your account. The voices are interchangeable and are also available in both services. So you can have consistent audio and video, yes. And to change the guest, you would just have to use or create a different avatar, yes. However, bear in mind that Notebook only outputs audio with the same two voices as seen in this video, so if you want to use Noteblook and to change one of the voices you'd have to either use a voice changer like in Eleven Labs bit.ly/HubJam-ElevenLabs , or transcribe the audio and have a text to voice service create the spoken audio. Hope that helps.
Thanks for watching and commenting. I explain in one point in the video that I had to cut it into three parts, due to being on Creator. Best regards, Jonathan.
Is there a way to load several sources in a notebook, ask a question and make a podcast out of that one answer (Note)? I tried making that answer a Note then its own source, then made only that source active, but could find no way to create a podcast out of that one source. I want to make several shorter podcasts from the sources, not one long podcast. If NotebookLM was able to do this, it would be a winner.
Thanks for watching and commenting. Well if that didn't work then I would suggest taking the content of the note you want to make the podcast out of, and creating a a new notebook and adding the note as the source, then generating the podcast out of that..?
I'm extremely fascinated by this concept my curiosity compels me to ask approximately and I know there's many steps involved but approximately how long does it take one to create a podcast from start to finish
Thanks for your comment. The time it takes depends how you do it, but if you follow the steps in this video, it's pretty quick. If you have a source prepared you could have it done in an hour but it might take two. Also depends if Notebook generates a short or a long audio. Hope that helps.
Hi, do you mean the recording of me? It's just a 25-70mm kit lens on a Sony a7s, personally I don't find it that sharp, and also Camtasia, which I use for screen recording and editing, I find compresses the image too much. The screen is recorded at 2k. Hope that helps.
yes this is true. I try to keep the Notebook audio to 10m (you can't control it, but supplying a shorter source info, and redoing it if it comes out too long normally gets it). The manual editing part is a pain, but if the video's ten mins and you just get into it, it's not that bad. Overall, it's much quicker than having to set up a recording of two real people, then do the actual recording which would always require editing out the retakes etc. Thanks for watching.
@@Jonathans-HubJam This is the era of AI automation...somebody has a tool that could do this drudgework for us. You can load the transcription and both characters will be mouthing the words of each of them....find a way to cut away from each one of them while the other is talking....surely some sort of AI ought to be able insert some sort of placeholder image of the other person just smiling and nodding. AI can add completely new details to an image, replace backgrounds, replace clouds etc etc....I am sure it can replace the clips where the person is speaking the other persons words with something else. I just don't know how to do it, myself. 😄
@@bobjones7274 I am sure this will happen. I don't know of it actually having been done yet. I'm sure some coder working with ai could do it. I am waiting for the day ai can edit my videos personally, then I'd be able to do one every day.
Awesome tutorial. Would love to chat if you're looking for a NotebookLM style podcast but with more creative control (choose the voices, edit script, more source types, etc.) But love the channel! - Pierson Jellypod, Founder
Glad you liked the tutorial! I just had a look at Jellypod - I think it looks very useful and people are already asking for more control than what Notebook provides. I notice you can't actually sign up for it yet.. when do you think it'll be ready? Regards, Jonathan.
Also, when you say two heads - do you mean two heads in the same frame/video? If so, how do you do that, I have had a look and don't see that function although I don't have all the functions as I don't have a Windows PC.
hmmm...🤔🙂 Its certainly cool tech...But...not sure...I think taking others content and using it verbatim ie taking from article, video or podcast and converting it into a conversation by one, two, three avatars etc (or by real people for that matter, such as reading a book to camera) would be copyright infringement unless its only a snippet or the content is radically altered ie actually writing different words, or a new thesis POV etc. For example if I took you video above and did that I don't think thats legal without asking permission plus surely morally dubious practise🤫
Hi and thanks for commenting. As another viewer has said, what NotebookLM does is to use generative ai to create a new dialogue / conversation that discusses the source content. So the words are actually different and don't just read out the source content at all. They only refer to it. Also, some website do have protection placed on their content, and that will be detected by NotebookLM and won't be processed. Also, one of the very useful functions of NotebookLM is to input multiple sources on the same topic and be able to chat with the LLM and ask it questions about the source content, and have it extract key information, organise it, etc. Hope that helps.
Hi and thanks for your comment. Which part do you need help with? Are you trying on HeyGen to make a video avatar of yourself? I have a video all about that - it's here: th-cam.com/video/TtKdgVMC1n0/w-d-xo.html Let me know if you still need help. Best regards.
➡ *Get HeyGen* heygen.com/?sid=rewardful&via=jonathan-hubjam I really hope this video was of value to you. Any questions please leave them below.
maybe synaesthesia for free gen?
NotebookLM is truly an extremely useful tool!
It certainly is!
This is perhaps the most useful tutorial on AI video generation I have watched the past 6 months. Thank you Jonathan. The world needs people like you. Keep it up. I really need this information. Subscribed and liked.
Wow, that is quite a compliment! That really is great to hear. Very motivating indeed. I always try to do my best. For some reason this video has done better than all my previous ones, but I always do things to this level. So glad to hear it was useful to you.
Soon I'll also be doing a video on making the podcast with your own avatar in it, as well as a new tool I found that allows you to change the voice etc. Best regards and thanks for watching, commenting, and subscribing!
Wow, that video looks super real! Great job on the tutorial, keep it up!
Thanks for watching and commenting. So glad you liked it! Best regards.
Great video. You explained the process perfectly.
Thanks for watching and commenting! So glad you found it to be useful. Best regards.
@@Jonathans-HubJam Brilliant work, but I wonder, why there is not a single tool in this AI era, that can separate male pitch from a female pitch? I think this this is easy but no body bother to create the one such,
@@librakhan25 Yes, surely those that know about ai voice technology could do that easily! I hope they do it soon.. Glad you liked the video by the way. Best regards, Jonathan.
@@Jonathans-HubJam I have already tried your method on HeyGen, I'm going to attempt Hedra now.
Been doing the audio bit for a while now and have a channel on Spotify and use this also for my work. With a little creativity, one can take this quite a long way. And I guess, this is only the beginning. Thanks for enhancing this with the video side of it.
Thanks for watching and commenting. Sounds great - how is the content doing on Spotify? Yes, this has only just begun, for sure! Regards.
@@Jonathans-HubJam I'm still very much in beta mode and taking it slowly. If you want, I'll DM the link. Regards
@@frankinbrida oh yes! please do share it!
That was so well explained! Thank you 🙏
Thanks for watching and commenting. So glad to hear you found it to be well explained! Regards.
Hi Jonathan, I've been experimenting with it for two weeks now and yes, it's been well received in my LEGO niche and my local language. Unfortunately, the tool still has extreme accents that are difficult to get rid of in German. But it's fascinating beyond belief.
Thanks for watching and commenting! Oh very interesting, LEGO, eh? So you're making videos from the Notebook dialogue? And are you using HeyGen for the voices? Regards.
@@Jonathans-HubJam Hello Jonathan, no, I use other people's videos that have information about LEGO products that we don't have here in Germany yet. That means I take the transcription, translate it into German and paste it into Notebook and have a dialogue/podcast produced. It usually works the first time. The soundtrack is then edited and cut accordingly. Then it is published.
Oh I see, that's a use case I'd never have thought of! I haven't tried it in other languages yet. I speak fluent Spanish so want to give that a try. By the way I am very soon presenting a new couple of tools you might be interested in - one that makes a complete two person talking video directly from a podcast audio,, and another which is an ai podcast creation tool that gives you complete customisation and control over voices and content. Look out for it! Best regards.
@@Jonathans-HubJam Yes, here I am once again way ahead in the LEGO niche. Some people in the bubble are trying to use Suno to create background music or chatgtp teasers. But I claim to be putting in more effort and experimenting here.
Now, I'm excited to see you produce a complete AI-generated video with the audio podcast! You have my subscription anyway. I would of course be delighted to get the famous click back from you. Maybe you can even use one of the podcast videos as an example to show that others are already trying it. Because I and my small team are already working on making the image avatars speak.
@@GenialesTuning Subscribed! Also my next video (talking about my viewer's suggestions for improvements to the process) will introduce this service of making a video from a notebook podcast (It's not available for people to use yet, but should be any day now), then I'll do a more in-depth video on it.
This is madness, amazing video Mr. Also you can change the voice with eleven labs using the switcher voice module.
Great tutorial. The world needs this
Thanks for watching and commenting. So glad to hear you liked my tutorial. Best regards, Jonathan.
I've used Hedra a few times to try and make a singing avatar for my Suno songs, using their provided "isolated vocals" (which aren't the cleanest, but are decent). It's never worked out very well, with the movements and lip sync being pretty low-quality. From your video, it looks like it handles clean speech much better than singing, though. It's an impressive result, for sure.
Thanks for watching and commenting. That's very interesting to know because I haven't tried Hedra with singing. Best regards.
Chopping up the audio to separate 2 speakers sounds like a labour intensive task.
Perhaps using Speaker Diarization tools, for example, from Assembly AI would help speed things up.
Thanks for watching and commenting. Thanks so much for your suggestion! I will look into it! Regards, Jonathan.
Wow this comment really help me a long way thanks,, been searching for days
@@JerseyBlack-xr6qg so glad it helped you!
Hi, talkingavatar has a Podcast Avatar feature, which could automatically diarizing the speaker, matching audio segments to the speaker, and ensuring lip-sync Podcast Video. With this new feature, you can generate a Podcast Video directly from NotebookLM audio with just one click, which is much more convenient.
Yes absolutely, you already informed me of that amazing possibility, and I am right now finishing up a video about it! Best regards.
Earned yourself a sub, comment and like, great video. Thank you
That is fantastic! So glad you enjoyed the video. Best regards.
Wonderful video. Thank you so much! Please make a video on how to use your own avatar.
Thanks for watching and so glad you liked the video! That video is in the pipeline!
@ Eagerly waiting!
I've got a handful of Notebook and podcast related videos coming, actually..
Use Voice Split Ai to split voices in different audio files.
Yes I'm literally right now doing a video on that. Thanks for your suggestion!
Thank you Jonathan for this wonderful class! I loved it! I wonder if there is a way to generate the same video with different languages or generate the subtitles in foreign languages so you can reach much more audience. Thank you again. At this very moment my choice would be the Hedra mode. 🙏
Thanks for watching and thanks for the compliment. So glad to hear you loved it! You could transcribe the audio from Notebook by using Descript, then use the transcription to generate speech with a text-to-speech-service, such as Eleven Labs, then you'd make the avatar video as in my video.
Alternatively you could just use Eleven Labs "Dubbing Studio" which takes any video or audio and allows you to dub it into a different language, so you could either make the original language version avatar video first, then dub it, or you could start with the audio from NotebookLM and import it into the Dubbing Studio to make other videos from that in other languages.
I think Eleven Labs' dubbing studio is still quite unknown but it's really powerful, check it out: bit.ly/HubJam-ElevenLabs
@@Jonathans-HubJam So kind of you answering me. Thank you for your response. At this very moment I am at my last PhD year, and my thesis is about AI in Education and Teacher Education. So, I am learning and intend to work with that, helping others to improve their teaching and learning processes. Thank you again! A big Brazilian hug! 🙏
@@OliveirosDiasJr it's a pleasure! Good luck with everything.
Thanks for the info. I'll give it a try. However, the audio editing looks like a pain but I'm up for the challenge 😂
Yeah it is a bit mind-numbing, but not too bad once you get into the flow. The trick is to try to get Notebook to output not too long - ie 20mins would take much longer to edit than 10, so try to keep the source as concise as possible. Regards.
@Jonathans-HubJam agree! However, I'm using NLM for doing book reviews, mostly about Food, Cooking and the Science behind it. So the podcasts will be on average, 15 - 20 minutes. 🙂 However, I'm ready to put in the grind 😁.
@@LarryFournillier I'm sure that very soon something new will come out that can do the dirty work for us! AI agents I presume. Regards.
@@Jonathans-HubJam yep, fingers are crossed.🤞🏾🙂
If you're interested in a little bit more creative control in the editing/creation process, would love for you to check us out! NotebookLM for Creators
Thanks for the video.
Is it possible to have the voice saved and the avatar (host of podcast) saved too?
To create consistent audio and video with the same Avatar (host of podcast) but different guests?
I was sure I'd answered this, but looks like I didn't send.. if I understand your question right, the avatars in Heygen are always available in their library, and the Hedra avatars are available in your account. The voices are interchangeable and are also available in both services. So you can have consistent audio and video, yes.
And to change the guest, you would just have to use or create a different avatar, yes.
However, bear in mind that Notebook only outputs audio with the same two voices as seen in this video, so if you want to use Noteblook and to change one of the voices you'd have to either use a voice changer like in Eleven Labs bit.ly/HubJam-ElevenLabs
, or transcribe the audio and have a text to voice service create the spoken audio.
Hope that helps.
Great video…but wasn’t clear to me was how did you manage to load the 10 min audio file to HeyGen using the creator plan.
Thanks for watching and commenting. I explain in one point in the video that I had to cut it into three parts, due to being on Creator. Best regards, Jonathan.
Is there a way to load several sources in a notebook, ask a question and make a podcast out of that one answer (Note)? I tried making that answer a Note then its own source, then made only that source active, but could find no way to create a podcast out of that one source. I want to make several shorter podcasts from the sources, not one long podcast.
If NotebookLM was able to do this, it would be a winner.
Thanks for watching and commenting. Well if that didn't work then I would suggest taking the content of the note you want to make the podcast out of, and creating a a new notebook and adding the note as the source, then generating the podcast out of that..?
@@Jonathans-HubJamyes thanks. That’s what I ended up doing.
I will try it in AKool AI and let you know how it goes.
Hi and thank for your comment.. hmmm AKool AI? I'll look that up, but please do let me know how it goes. Thanks.
did you try multicam in capcut? i think it can be done using it, much easier!!! thanks for the video, btw. 😍 cheers from 🇧🇷
Thanks for watching and commenting and glad you liked the video! No I didn't try multi-cam, didn't think of it - will have a look at that! Thanks.
Hey, I checked the multi-cam method out and just made a video about it - thanks for the suggestion! th-cam.com/video/EXA5jfoiK6M/w-d-xo.html
I'm extremely fascinated by this concept my curiosity compels me to ask approximately and I know there's many steps involved but approximately how long does it take one to create a podcast from start to finish
Thanks for your comment. The time it takes depends how you do it, but if you follow the steps in this video, it's pretty quick. If you have a source prepared you could have it done in an hour but it might take two. Also depends if Notebook generates a short or a long audio. Hope that helps.
I would be interested in more details.
Hi thanks for your comment. What would you like to know?
Please make a video on topic "Ai History TH-cam faceless channel" like ai podcast TH-cam channel.
Thanks for your comment. Yes, that is something I wanted to do. It's on my list, but might be a while.
Wow...which camera do you use ? So crisp !
Hi, do you mean the recording of me? It's just a 25-70mm kit lens on a Sony a7s, personally I don't find it that sharp, and also Camtasia, which I use for screen recording and editing, I find compresses the image too much. The screen is recorded at 2k. Hope that helps.
Thx
A pleasure. Thanks for watching!
There has got to be a way to speed up the process. You basically have to spend more time than the podcast lasts in order to do all of the editing etc.
yes this is true. I try to keep the Notebook audio to 10m (you can't control it, but supplying a shorter source info, and redoing it if it comes out too long normally gets it). The manual editing part is a pain, but if the video's ten mins and you just get into it, it's not that bad.
Overall, it's much quicker than having to set up a recording of two real people, then do the actual recording which would always require editing out the retakes etc.
Thanks for watching.
@@Jonathans-HubJam This is the era of AI automation...somebody has a tool that could do this drudgework for us. You can load the transcription and both characters will be mouthing the words of each of them....find a way to cut away from each one of them while the other is talking....surely some sort of AI ought to be able insert some sort of placeholder image of the other person just smiling and nodding. AI can add completely new details to an image, replace backgrounds, replace clouds etc etc....I am sure it can replace the clips where the person is speaking the other persons words with something else. I just don't know how to do it, myself. 😄
maybe something like this can help you edit your video: th-cam.com/video/1C1mNryWX_M/w-d-xo.html
@@bobjones7274 I am sure this will happen. I don't know of it actually having been done yet. I'm sure some coder working with ai could do it. I am waiting for the day ai can edit my videos personally, then I'd be able to do one every day.
Yeah it’s called work
Awesome tutorial. Would love to chat if you're looking for a NotebookLM style podcast but with more creative control (choose the voices, edit script, more source types, etc.)
But love the channel!
- Pierson
Jellypod, Founder
Glad you liked the tutorial! I just had a look at Jellypod - I think it looks very useful and people are already asking for more control than what Notebook provides. I notice you can't actually sign up for it yet.. when do you think it'll be ready? Regards, Jonathan.
@@Jonathans-HubJam We're on a waitlist but letting people off via email, DM, or joining the Discord!
Public signups coming later this month.
Idea. Use Eleven Labs to change to new voices.
Yes! Will do a video on that
@@Jonathans-HubJam Please do
TalkingAvatar will do it with two heads and is free
Thanks for the suggestion! Will look into it. Regards.
Could you share the link? Thank you
www.talkingavatar.ai/
Also, when you say two heads - do you mean two heads in the same frame/video? If so, how do you do that, I have had a look and don't see that function although I don't have all the functions as I don't have a Windows PC.
actually I've seen it now - yes they can do two heads on the same video - amazing. Thanks for recommending.
hmmm...🤔🙂 Its certainly cool tech...But...not sure...I think taking others content and using it verbatim ie taking from article, video or podcast and converting it into a conversation by one, two, three avatars etc (or by real people for that matter, such as reading a book to camera) would be copyright infringement unless its only a snippet or the content is radically altered ie actually writing different words, or a new thesis POV etc. For example if I took you video above and did that I don't think thats legal without asking permission plus surely morally dubious practise🤫
In this video, and my own understanding, the NotebookLM voices are commenting on the source material, not reading it (or quoting AFAIK).
Hi and thanks for commenting. As another viewer has said, what NotebookLM does is to use generative ai to create a new dialogue / conversation that discusses the source content. So the words are actually different and don't just read out the source content at all. They only refer to it. Also, some website do have protection placed on their content, and that will be detected by NotebookLM and won't be processed. Also, one of the very useful functions of NotebookLM is to input multiple sources on the same topic and be able to chat with the LLM and ask it questions about the source content, and have it extract key information, organise it, etc. Hope that helps.
not "perfectly", but it's getting close
Please help me to create my video avatar
Hi and thanks for your comment. Which part do you need help with? Are you trying on HeyGen to make a video avatar of yourself?
I have a video all about that - it's here: th-cam.com/video/TtKdgVMC1n0/w-d-xo.html
Let me know if you still need help. Best regards.
Couldnt think of easier names then Migumi and Vernon, John and Jane, how bout Adam and Eve.... Migumi and Vernon 🤦
Ha ha! 😄😆 Yeah didn't thinkn of that! Vernon is the actual name of that avatar though. And Megumi is the name of a Japanese friend of mine. Hey ho..