{"event_embed_url":"https://heygen.wistia.com/embed/live/events/0o7f21hj66/audience","hashed_id":"0o7f21hj66","media":{"accountId":2397428,"accountHashedId":"69y4jlt51m","accountKey":"wistia-production_2397428","analyticsHost":"https://distillery.wistia.net","formsHost":"https://app.wistia.com","formEventsApi":"/form-stream/1.0.0","aspectRatio":1.7777777777777777,"assets":[{"is_enhanced":false,"is_composite_layer":false,"type":"still_image","slug":"still_image_1200x720","display_name":"Thumbnail Image","details":{},"width":1200,"height":720,"size":685617,"bitrate":0,"public":true,"status":2,"progress":1.0,"metadata":{"av_stream_metadata":"{\"FastStart\":false,\"Audio\":{},\"Video\":{\"codec\":\"png\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"gbr\",\"colorTransfer\":\"iec61966-2-1\",\"height\":720,\"pixelFormat\":\"rgba\",\"rFrameRate\":\"25/1\",\"width\":1200}}"},"url":"https://embed-ssl.wistia.com/deliveries/67545b31b3a76064cb84ef300fb71154.bin","created_at":1792440327},{"is_enhanced":false,"is_composite_layer":false,"type":"original","slug":"original","display_name":"Original File","details":{},"width":1920,"height":1080,"ext":"","size":1007422893,"bitrate":3279,"public":true,"status":2,"progress":1.0,"metadata":{"av_stream_metadata":"{\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.269932,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"3216328\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":1080,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":1920}}","max_bitrate":5098880,"average_bitrate":3357698,"early_max_bitrate":4142856,"time_to_ready":47.496815885},"url":"https://embed-ssl.wistia.com/deliveries/5856f9f57ee3610532589c3f999356ac.bin","created_at":1793122858},{"is_enhanced":false,"is_composite_layer":false,"type":"iphone_video","slug":"mp4_h264_222k","display_name":"360p","details":{},"container":"mp4","codec":"h264","width":640,"height":360,"ext":"mp4","size":68378336,"bitrate":222,"public":true,"status":2,"progress":1.0,"metadata":{"max_bitrate":413392,"average_bitrate":227901,"early_max_bitrate":267944,"av_stream_metadata":"{\"FastStart\":true,\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.274626,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"93355\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":360,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":640}}","built_by":"render-pipeline","time_to_ready":58.236401167},"url":"https://embed-ssl.wistia.com/deliveries/122d09710df6763c6a06f15e93e5c02152975337.bin","created_at":1793122904,"segment_duration":3,"opt_vbitrate":1200},{"is_enhanced":false,"is_composite_layer":false,"type":"mp4_video","slug":"mp4_h264_179k","display_name":"224p","details":{},"container":"mp4","codec":"h264","width":400,"height":224,"ext":"mp4","size":55243335,"bitrate":179,"public":true,"status":2,"progress":1.0,"metadata":{"built_by":"gen3-split-n-stitch","max_bitrate":268648,"average_bitrate":184123,"early_max_bitrate":202280,"av_stream_metadata":"{\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.274626,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"49610\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":224,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":400}}","time_to_ready":78.522601257},"url":"https://embed-ssl.wistia.com/deliveries/8ef8a53f216721c5059e3b206adc591cbbacbe0b.bin","created_at":1793122904,"segment_duration":3,"opt_vbitrate":300},{"is_enhanced":false,"is_composite_layer":false,"type":"md_mp4_video","slug":"mp4_h264_293k","display_name":"540p","details":{},"container":"mp4","codec":"h264","width":960,"height":540,"ext":"mp4","size":90182001,"bitrate":293,"public":true,"status":2,"progress":1.0,"metadata":{"max_bitrate":680856,"average_bitrate":300572,"early_max_bitrate":372960,"av_stream_metadata":"{\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.274626,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"166017\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":540,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":960}}","built_by":"gen3-split-n-stitch","time_to_ready":132.760812922},"url":"https://embed-ssl.wistia.com/deliveries/c6123d1c1201554037d70bea3f2625976f902662.bin","created_at":1793122904,"segment_duration":3,"opt_vbitrate":1800},{"is_enhanced":false,"is_composite_layer":false,"type":"hd_mp4_video","slug":"mp4_h264_374k","display_name":"720p","details":{},"container":"mp4","codec":"h264","width":1280,"height":720,"ext":"mp4","size":115096966,"bitrate":374,"public":true,"status":2,"progress":1.0,"metadata":{"max_bitrate":995408,"average_bitrate":383612,"early_max_bitrate":494736,"av_stream_metadata":"{\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.274626,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"249058\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":720,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":1280}}","built_by":"gen3-split-n-stitch","time_to_ready":157.76977073},"url":"https://embed-ssl.wistia.com/deliveries/d33eb78819930cc8446a155b53e1ae82f8bac31a.bin","created_at":1793122904,"segment_duration":3,"opt_vbitrate":3750},{"is_enhanced":false,"is_composite_layer":false,"type":"hd_mp4_video","slug":"mp4_h264_585k","display_name":"1080p","details":{},"container":"mp4","codec":"h264","width":1920,"height":1080,"ext":"mp4","size":179878352,"bitrate":585,"public":true,"status":2,"progress":1.0,"metadata":{"built_by":"render-pipeline","max_bitrate":1928480,"average_bitrate":599525,"early_max_bitrate":840480,"av_stream_metadata":"{\"FastStart\":true,\"Audio\":{\"channels\":2,\"channelLayout\":\"stereo\",\"Codec ID\":\"mp4a-40-2\",\"codec\":\"aac\",\"codecType\":\"audio\",\"duration\":2400.274626,\"sampleFormat\":\"fltp\",\"sampleRate\":44100},\"Video\":{\"avgFrameRate\":\"30/1\",\"bitRate\":\"464975\",\"Codec ID\":\"avc1\",\"codec\":\"h264\",\"colorPrimaries\":\"bt709\",\"colorSpace\":\"bt709\",\"colorTransfer\":\"bt709\",\"duration\":2400.233333,\"height\":1080,\"pixelFormat\":\"yuv420p\",\"rFrameRate\":\"30/1\",\"width\":1920}}","time_to_ready":65.975290976},"url":"https://embed-ssl.wistia.com/deliveries/c3534f635e215e718b43bc6fcbac0bc8d5f5d2f4.bin","created_at":1793122904,"segment_duration":3,"opt_vbitrate":5625},{"is_enhanced":false,"is_composite_layer":false,"type":"storyboard","slug":"storyboard_2000x2240","display_name":"Storyboard","details":{},"width":2000,"height":2240,"ext":"jpg","size":656816,"bitrate":0,"public":true,"status":2,"progress":1.0,"metadata":{"frame_count":200,"aspect_ratio":1.7777778,"frame_height":112,"frame_width":200,"built_by":"render-pipeline","time_to_ready":45.173997857},"url":"https://embed-ssl.wistia.com/deliveries/ccd90d06ef2227b5f96b414c3d151f3f8f04b899.bin","created_at":1793122904},{"is_enhanced":false,"is_composite_layer":false,"type":"storyboard","slug":"storyboard_2000x2240","display_name":"Storyboard","details":{},"width":2000,"height":2240,"ext":"jpg","size":656816,"bitrate":0,"public":true,"status":2,"progress":1.0,"metadata":{"frame_count":200,"aspect_ratio":1.7777778,"frame_height":112,"frame_width":200,"built_by":"render-pipeline","time_to_ready":48.912474464},"url":"https://embed-ssl.wistia.com/deliveries/f1897afb593a3d5d00fb523f06bc825fb486a336.bin","created_at":1793122905},{"is_enhanced":false,"is_composite_layer":false,"type":"storyboard","slug":"storyboard_2000x2240","display_name":"Storyboard","details":{},"width":2000,"height":2240,"ext":"jpg","size":656816,"bitrate":0,"public":true,"status":2,"progress":1.0,"metadata":{"built_by":"render-pipeline","frame_count":200,"aspect_ratio":1.7777778,"frame_height":112,"frame_width":200,"time_to_ready":45.898702495},"url":"https://embed-ssl.wistia.com/deliveries/4af1b3609b23279b61d50deed5c9b2bdbcb3d72d.bin","created_at":1793122905}],"branding":false,"createdAt":1792439954,"distilleryUrl":"https://distillery.wistia.com/x","duration":2400.27,"enableCustomerLogo":true,"firstEmbedForAccount":false,"firstShareForAccount":false,"availableTranscripts":[{"bcp47LanguageTag":"en","familyName":"English","familyNativeName":"English","hasCaptions":true,"hasVideoStream":false,"name":"English","nativeName":"English","wistiaLanguageCode":"eng","language":"eng","ietf_language_tag":"eng","alpha3_terminologic":"eng","alpha3_bibliographic":"eng","iso639_2_language_code":"en","english_name":"English","native_name":"English","right_to_left":false,"alpha3Bibliographic":"eng","alpha3Terminologic":"eng","ietfLanguageTag":"eng","iso6392LanguageCode":"en","genericName":"English","genericNativeName":"English"}],"hashedId":"l096j2cxpp","mediaId":154206791,"mediaKey":"wistia-production_154206791","mediaType":"Video","name":"Live demo: HeyGen Voice, the #1 voice model","preloadPreference":null,"progress":1.0,"protected":false,"projectId":10877572,"seoDescription":"a Sample live event video","showAbout":true,"status":2,"type":"Video","originalIsEligibleForDirectPlayback":true,"originalFileStillAsset":{"url":"https://embed-ssl.wistia.com/deliveries/5856f9f57ee3610532589c3f999356ac.jpg?video_still_time=1200","width":1920,"height":1080,"type":"still_image","status":2,"public":true,"progress":1.0,"display_name":"Thumbnail image","details":{},"metadata":{}},"playableWithoutInstantHls":true,"stats":{"loadCount":349,"playCount":204,"uniqueLoadCount":247,"uniquePlayCount":173,"averageEngagement":0.0},"trackingTransmitInterval":80,"liveStreamEventDetails":{"id":436161},"integrations":{},"tagManager":true,"captions":[{"language":"eng","text":"Hello, everyone. It's great to see you all. Hey, Adam. Thanks for joining today.\n\nHey, Excited to be here.\n\nI'm joining from San Francisco. I'm Audrey on our community team at HeyGen. We've got Adam here to talk about HeyGen voice today.\n\nAnd before we get started, a couple housekeeping notes.\n\nIf you've got questions, please drop them in the q and a tab.\n\nWe'd love to also know where you're joining from. Feel free to drop in the chat where you're joining from in the world. I'm calling in from San Francisco today.\n\nWe've got a couple people helping out in the chat as well.\n\nAnd just a note that if you need any assistance, if you have questions that don't get don't get answered here today, you can already always reach out to Hagen support at support@Hagen.com.\n\nYou can also check out Hegen Academy.\n\nAnd we'd love to see you for a couple more sessions. We have a couple other live sessions happening this week. We've got a support office hours session. And we've got a fun HeGen Halloween webinar that will go into creating looks.\n\nIf you are brand new to HeGen, before we jump into it, HeGen is an AI video platform. It lets you create videos starring you, your voice, the way you move, so that you don't need to film yourself. You can be your best self on camera every day without ever having to film yourself again.\n\nSo if you are new to HeyGen, we've got a special offer for you for joining our webinar today. You can scan this QR code, and it'll give you your first month of creator free. This lets you create your avatar, start making videos, create your voice clone that Adam's gonna walk you through.\n\nSo definitely scan this. I'll leave this on the screen just for another minute before we jump into it. For anyone who is new, this will let you get started right away. So definitely encourage you to scan this so you can get started with your first month free.\n\nAwesome. Well, Adam, I will hand it over to you to get into everything Hegen Voice because I know there's a lot of exciting updates and things to share with everyone today.\n\nAwesome. Thank you, Audrey. I will share my screen. Can you see this? Alright.\n\nYes. It looks great.\n\nPerfect.\n\nCool. So, yeah, my name is Adam Halper. I'm a product manager at Hegen. I work on our avatar voice and video models. And today, very excited to talk to everyone about Hegen voice.\n\nSo what we'll cover today is, first, why do most AI voices sound off when you're using them in videos, what makes HeyGen Voice different, we'll go through live demos on how you can create your own voice on our app and API, and how you can also tune it so you nail the delivery and can fit any scene. So we'll try to go through quick as well so we save a lot of time for anyone's questions at the end.\n\nAnd starting with why do most AI voices sound off when you're using them in videos? The reason is is because most AI voices are not built for using them on video.\n\nMost AI voice companies are are building for conversational agents or narration, and the data they train on is largely from conversations or data that's being read.\n\nAnd so when you use those voice models, often it can sound monotone like it's being read, not presented or performed.\n\nAnd as a result, it can often sound monotone. So we see sounding monotone has been one of the most common issues that our users complain about. And when you try to offset that by adding expressiveness, it can often then not sound like you. So traditionally users often have to face either sound monotone or just sound like someone else. And then additionally, when you're using a voice and video, there's this added constraint where it should sound like it matches your setting because I sound differently when I'm indoors versus when I'm outdoors. And when the voice, the audio doesn't match the setting, there can be that disconnect where it just feels a little less realistic, a bit uncanny.\n\nSo with Hegen Voice, we set out to solve these problems. And the way we did it is first, we trained our Hegen Voice model on video audio data.\n\nYou know, as a video company, we have a lot of experience with this, And the impact of this is when you use this voice in videos, it sounds natural, like you're really speaking on a video and expressive. It's built for performance.\n\nSecondly, we paid a really strong emphasis on likeness, meaning making sure your voice sounds like you. And one of the key ways we did this is we take a longer context, meaning more of your actual audio input when cloning your voice and generating your text to speech. Whereas most other model providers are focused on actually getting it as small as possible, allowing you to clone your voice off five, ten, fifteen seconds so that it's fast and easy.\n\nWe kind of went the other direction where we're taking in up to sixty seconds and with our PVC up to two hours. And what that allows is us to have a bigger range of your actual real voice, which means you can both sound expressive without having to make it up.\n\nWhen you just have five, ten seconds of data on your voice, you end up either sounding kind of flat and one dimensional or it has to make it up. But with HeyGen Voice, it does a good job of balancing both the expressiveness while still keeping it like you. And then finally, we built a really unique feature, which is our video adaptive setting. And when you turn this setting on, what it does is it adapts the audio to fit the look in which you're creating the video in. So it fits your environment. And under the hood, this means when you are actually generating the audio, we pass in not just your script but also the scene, like the image, the look, your avatar's look from where you're creating from, and so that audio comes out matching that scene. So if I'm outside in the park, it would sound like I'm speaking outside, might potentially have some rustling leaves, some bird chirping, really making it seem like I'm realistically speaking from that setting.\n\nAnd since we've launched this model, we're seeing really strong reception both on our platform, it's the most preferred engine. When users create voices on Heijen, we always offer all the best options, let users choose often in a blind way. And we see users are preferring Heijen Voice over Cartesia, over Eleven Labs.\n\nAlso, in independent evaluations, artificial analysis is probably the leading independent benchmark. They evaluate models across various dimensions and they find HeGen Voice and HeGen Voice Turbo are the top one and two models for controlled voice in the world right now. So all this is to say, if you haven't tried out HeGen Voice, we'd definitely recommend that you give it a try.\n\nHere are a couple of demonstrations. I might need to reshare. Audrey, can you call out if you can't hear this?\n\nI grew up three hours from the coast, which means I spent most of my childhood being told the beach was a special occasion. Now I live ten minutes from one.\n\nSo I was with 11 laps before. Good voice.\n\nI grew up three hours from the coast, which means I spent most of my childhood being told the beach was a special occasion. Now I live ten minutes from one.\n\nSo this is with that video adaptive setting on. Now as you can hear, it clearly sounds like he's outside. You can kind of hear the waves you know, coming in and out. So just much more of an immersive vibe if you really want that realism. Another example.\n\nI started climbing because I was afraid of heights. That sounds like a joke, but it isn't.\n\nI figured the only way to And then with the video adaptive on.\n\nI started climbing because I was afraid of heights. That sounds like a joke, but it isn't. I figured the only way to dismantle a fear was to walk directly into it repeatedly.\n\nAgain, can kind of just hear that ambient wind sounds in the background. And it's all subtle, but that subtle connection between the audio and the visual can really help it cross an uncanny valley and just make it feel a lot more realistic to people who are watching.\n\nOne more example.\n\nI have worked in this building for twenty eight years. I have seen it change. The machines, the owner...\n\nI have worked in this building for twenty eight years. I have seen it change the machines, the ownership, the shifts, everything.\n\nI So, yeah.\n\nSo in each case, you know, it sees the actual environment, then it's adding those subtle ambient sounds. And this is the setting which I'll show in our live demo. You can control the strength of it as well.\n\nLast demo is of me on a beach. And here I'll show the adaptive off and also with it on.\n\nFirst, maybe I'll play with it on.\n\nNow I live ten minutes one.\n\nSo as you can hear, the adaptive off still actually it sounds great. It had that clean studio quality sound. And when the adaptive's off, HeyGen Voice will go for a very clean studio quality sound, which for a lot of videos, the majority of videos, you're probably better off with that adaptive offsetting. But when you are in a specific situation where it's on a beach, in a park, somewhere where it does merit having those environmental sounds and you really want that added realism, the adaptive setting works great there.\n\nAwesome. So now I will quickly go and show how you can actually create your voice both on the app and then will show on our API.\n\nLet me reshare.\n\nCool. So first, I will show how can you create your voice on the Hegen app. So first, you go to avatars.\n\nAnd it's worth pointing out, whenever you create a new avatar, you will also get a new voice clone through that process.\n\nSo you haven't already created your avatar, I'd recommend tap new avatar and go through this flow of creating for myself.\n\nBut let's say, you know, you already have an avatar, you want a new voice, or just wanna create a standalone voice clone, which you can use for your videos, but also just for any voice need.\n\nYou can go to avatar, the voices tab, and then tap create your HeyGen voice.\n\nSo here you have a few options.\n\nFirst, worth calling out, if you already have good audio, and let's say I've already been recording videos the traditional way, posting them to YouTube, that can often be great audio to use to create your voice clone, where a video that you've previously created is a script that you've actually used, So it covers you talking naturally. The fact that you've published it means you probably liked how it sounds. So if you're using a already published video as your training to create your new voice clone or, you know, already published audio, then it's likely you'll like the voice clone that comes out of that.\n\nBut if you don't already have, you know, existing audio to use, you can record audio here.\n\nAnd then you have a few options. You can record off your smartphone or you can record directly on the computer. I would suggest high quality equipment, the better. So a smartphone works great.\n\nIf you have a newer laptop with a good laptop mic, that also works. Or if you have a professional mic you can plug into your laptop, that's perfect.\n\nAnd then this script is really designed to help you be expressive, to enunciate clearly, to vary your tone a bit so that it doesn't come out at one pace and tone the entire time. That said, we actually see the very best voice clones often come from when you are using a script that is very natural to you and personal to the videos that you're actually creating. Which means it can often be good if you want to write your own script and that script should be a script that you'd actually use in a video. Ideally, it should also be expressive, make sure you're enunciating clearly.\n\nBut if it covers a topic you actually talk about, you're more likely to speak naturally in your own cadence.\n\nAlso, it's good if it can include proper nouns like your business name, your product name, as if it hears it, it's more likely to pronounce it correctly when it encounters it, and also any accent markers that you have. So if there's specific words that you pronounce uniquely, like if I say like whata, then that's a good word to include here as, you know, it'll get my accent right when it encounters those situations going forward.\n\nSo I can tap start recording. Now I'd be reading the script.\n\nAt the end, I could, you know, tap stop recording. For the sake of this demo, I'll just skip past to this is a voice I created, you know, last night.\n\nAnd now I can go and use this voice to create videos with across any surface on Heijen. Maybe I will show how you can do it on the AI Studio.\n\nSo I'll tap new video.\n\nAnd then on the side panel, you see voice, and every avatar by default has a voice set to it.\n\nFor the sake of this demo, I'll just find the...\n\nThat common clear voice. So this is the voice we are just looking at. So I can select it, and now I can just write my script and generate. And it's that easy.\n\nI'm now using my voice I just cloned with my avatar, and and that's really all you need to do to use Hedgehog Voice to create videos. If you do wanna fine tune it, we do have a bunch of settings you can do to really nail it. I'll go over now. First, you have the adaptive to luck setting.\n\nThis is what I was speaking about earlier.\n\nWhen this is on, the voice should adapt to this scene. So it'll sound like I'm outside, you might hear the birds chirping, whereas when this is off, it'll have that studio quality sound that you heard earlier.\n\nAnd this similarity setting dictates how much it adapts to the scene. So when it's all the way on closer to original, it will not adapt to the scene at all. It'll really sound closer to your audio recording that you provided.\n\nWhen similarity is all the way up more adaptive, it'll really adapt to the scene, so you'll really hear the background sounds. It'll really sound like you are outside in this case.\n\nAnd again, by default, when the adaptive setting is off, it essentially opts for, like, a clean studio audio sound. So if you have this setting off, making it more adaptive will make it sound more like a clean studio audio, whereas making it less adaptive will make it sound closer to your original audio. So if you want that really clean studio audio sound, can keep it as more adaptive and just leave the setting off.\n\nWe also have an expressiveness setting. This is great for adding energy to your voice. So if it does sound monotone, increase this expressiveness. And this can also vary depending on the script. Like, if you have a script where you're announcing a very exciting win or selling something, you probably wanna turn it up. If you have, like, a training course, you might turn it down.\n\nAll of that said, this model is really built to adapt to the script.\n\nBased on the script you give it, it will sound more excited for something that's more excited. It'll sound calm when you have calm lines. So this is all good to help further guide it, but you shouldn't feel like it's all on you. For the most part, the model adapts based on the script you give it, and the goal is is that it should, you know, speak it naturally, but you always have these for added control.\n\nAdditionally, you have a tone control. So if you do wanna give it a specific tone, make it more excited, keep it calm, you can set that here.\n\nSo that is how you use Hegen voice.\n\nIt's available anywhere you create videos on Hegen. You can use your Hegen voices.\n\nI'll also just quickly show we have a new creative studio, and this allows you to create audio only tracks. So let's say I actually don't need to create it for a video purpose. I just have, like, a voice over I wanna, you know, create over, you know, something or even a podcast.\n\nHere, you can just add audio, and it'll generate that audio like this super fast.\n\nSo if you just need to generate audio, Creative Studio Voice is a great place to do that.\n\nAwesome. Now I will show how you can also use this on our API. And let me quickly reshare my screen.\n\nSo if you haven't checked it out yet, we recently are launching a new API platform where you can use all our models directly, you know, via the API. And this is really great in that you can build with your voice. If you're building your own applications, your own workflows, your own automations, you can really bring your voice anywhere with you.\n\nSo it's on platform.heygen.com. And I'll just start at the home section. Here, you can also see our our video models, our avatar models. But, you know, for the sake of this, I'll go to HeiGen Voice. And here, there is a playground where you can try it free.\n\nSo if you do wanna just, like, test out the different models, you can do that here. You'll notice all of the voices that are in your app, like, on my account on HeyGen, also exist here. So if I, you know, have the voice that I created on the app, I can also use it on the API. Like, here's that voice we just went over in the app.\n\nAnd also worth calling out, we are offering... So we have two endpoints for... Or two models for Hegen voice. We have Hegen Voice normal and Hegen Voice Turbo. Turbo is a better... Or it's a faster, cheaper version of the model, and we are offering it absolutely free for the next month on the API. So if you are building anything, if you do have a voice need or just wanna, you know, be able to iterate, you can use our HN Voice Turbo endpoint for free right now. All you need to do is create an API key, copy it, and then, you know, start using that turbo endpoint, which is here.\n\nAnd as a quick example, I will show how you can set up and actually create it, like, with an agent helping you.\n\nI have to reshare my screen one time.\n\nCool. Let me know if... Or James, Ajay, if you can't see this alright.\n\nBut what you can do is now you can actually use your voice on the API really without being technical. With agents, it can really... You can describe in your own words what you want to do with it, and it can help you.\n\nAnd on our playground, there's a copy prompt for your agent.\n\nSo you can copy this prompt. Previously, you can just create your API key here, generate one, and then all I'm gonna do is just... Into Codex, but you could do this with Claude, with GrockBot. I'm just pasting in that prompt, which is basically saying, I wanna use this voice, and it's gonna set it up for me. I could even say, like, use my Adam voice clone to generate clips saying, hello, everyone. Welcome to the webinar.\n\nSo right now, you know, I just copy pasted that prompt into Codex, and I can also just have it generate audio, you know, directly by using my API key.\n\nIf there's, you know, other pipelines, you really can have it do whatever you want here, whether that's integrating it into your own website, your own kind of pipelines, or just using this directly because you, you know, want to be able to have your agent help you clone the voice.\n\nSo, yeah, here is... It was that easy. I just pasted it, and now here's the generated clip. Hello, everyone. Welcome to the webinar.\n\nAwesome. So that is how you can create your voice on the app or also use it directly on the API.\n\nAgain, would highly recommend have the agent set it up for you, and it can really then help integrate it into whatever pipeline you want.\n\nNow I'll quickly go over just some tips for how to record a great voice. Went over this a bit before, but, again, if you already have an existing video or, you know, audio recording that you really like that you might have already published online, that can often make great source material for your clone.\n\nTo the extent that it's a real script that is representative of how you'll actually be using your voice clone, that makes it good input footage.\n\nBut if you don't already have existing footage you can use, how to record new footage, one, for equipment, smartphone works great. If you have a high quality mic, that's perfect.\n\nOr, you know, a newer mic on a, like, a a newer laptop, make sure your setting is pretty quiet, that there's not too much background noise. With the video adaptive setting, you can, you know, add the background noise, but you want your input audio to be clean.\n\nIf you are recording on an iPhone, would recommend use the Voice Memos app and turn on the lossless setting for better quality. And then if you are recording on a phone, generally the best technique is to hold it flat about six inches from your mouth to capture the best audio while minimizing the noise. And this is different from how you record your avatar. When you're recording your avatar, you have to be a bit farther back, you're using your gestures, and your mouth is necessarily farther from the mic. So sometimes we do find people get better quality when they separately record their voice clone.\n\nAs far as the delivery goes, you want to be very expressive but also speak in steady sentences and to avoid overly long pauses. Try to avoid pausing longer than a couple seconds during the script as that can introduce more instability when you're using it.\n\nFor the script, again, you can use R script, that works great, but the best script is a script that matches a script you'd actually use when creating with your avatar.\n\nIdeally, it includes basically something that allows you to speak very naturally in your own cadence and also includes proper nouns like your business, your product name, and if you have an accent, words that really capture your accent.\n\nAnd then, I guess, to be repetitive, be extra expressive. The number one piece of advice I generally give people is just when you're recording your voice clone, be a little bit more expressive than might be comfortable.\n\nThe cloning often flattens the tone a bit and adding a bit more energy than usual generally leads to people being happier with their outcome.\n\nI already went over the kind of different ways that you can control the settings, but actually one that I didn't go over so far is the custom pronunciation.\n\nI'll really quickly show how you can do that on our app. For custom pronunciation, we support both a brand glossary and inline pronunciation fixes.\n\nSo let's say...\n\nOh, I already fixed that one. Let's say our company name was Skydence.\n\nAnd let's say it wasn't pronouncing this correctly. All I need to do is highlight it. It'll... You'll see this pronunciation pop up, and I can add the pronunciation here. I can either spell it out or I could speak it out and then save it. And then that pronunciation, whenever, you know, it encounters that word, you know, in this video, it will fix that pronunciation to be exactly as I spelled it or spoke it out. And it also supports IPA.\n\nYou can have an agent help you get the IPA and paste that in as well.\n\nThe other option is you can use our brain glossary. This is a great option in that you can create a glossary and then add terms here.\n\nIn this case, I'll add guidance, the specific pronunciation I want for it. Now every time in all my videos, whenever it encounters, you know, any word in my brain glossary, it'll always pronounce it correctly according to my custom pronunciation.\n\nAnd you can just set which glossary you're using here. That glossary also is supported on our API.\n\nSo if you, you know, create your glossary here and then you're trying to use on the API, you can share that glossary across both.\n\nIn terms of common issues, if it doesn't sound like you, would recommend sliding that similarity slider closer to original.\n\nIf that doesn't work, rerecord for better high quality input. Often, this will mean higher quality mic, quieter room where it has a clean audio clip of how you actually sound.\n\nAlso, it's too short, we'd recommend recording at least sixty seconds. If it sounds flat, you can raise the expressiveness slider or choose one of the tones.\n\nAnd if that doesn't work, again, would recommend to rerecord your sample and make sure to add extra energy, be extra expressive. All of these settings do help, but something always holds true, which is high quality input equals high quality output. And if the input is really low quality, which means it's hard to actually... There's lots of background noise, scratchiness, there's not enough of it, it sounds monotone, then it's hard to turn that into good audio.\n\nSo I would always recommend try fixing it with the settings first, but if that doesn't work, re record.\n\nIf it doesn't match the video, turn on the adaptive setting, and if a word is mispronounced, you can fix it in line or with the brain glossary.\n\nCouple of constraints to call out.\n\nWe are working on making Heijen work great for all languages. Right now it officially supports 39. We see that it's strongest in English, also French, Spanish, Korean, Japanese, Chinese, German, Russian. These are languages where we see users significantly prefer Heijen Voice and are side by side. Not to say it doesn't support other languages, we just haven't officially evaluated all languages.\n\nSo regardless of your language, would recommend you try it out and we will be expanding these over time. I would love your feedback as this also helps inform our priorities of which languages to really focus on.\n\nGranular delivery controls is something coming soon. Right now, you have that scene level tags where you can make it friendly, calm, etcetera, but it doesn't have that inline word level tags that some other models offer where you can have it emphasize a specific word or have it laugh at a certain point.\n\nAll that is coming very soon though, so stay tuned.\n\nFinally, sometimes with the instant voice clone, you won't always nail your perfect likeness and this is where our professional voice clone comes in. The instant voice clone really passes in sixty seconds of audio whereas our professional voice clone takes a minimum of twenty minutes and up to two hours. So it has a real range of how you sound and that allows it to really nail your pronunciations, your cadence, your tone, and do it in a very consistent way where sometimes IVCs can nail it, sometimes it might sound a bit off, but if you want something that sounds like you and sounds like you every time, I would recommend trying out our professional voice clone. It is like a bit of an investment, but the advantage is once you create it once and you like it, you basically have this asset you can use for life that saves you a lot of time.\n\nAnd on the professional voice clone, I went through this, so mostly we'll just show the... These quick demos.\n\nAgain, it's the way to get the most accurate clone of your voice, really nails your likeness, does require a minimum of twenty minutes of training, and it is a paid add on.\n\nIt's currently $29 a month, but we find... Or if if you do wanna try it, we will be offering it for people free as well.\n\nSorry.\n\nSorry. Sharing some background noise on mine.\n\nCool. So let me reshare real quick.\n\nYou guys...\n\nYeah. This is, no. This is my screen right now, so you can take over from sharing.\n\nBack on the slides, Adam.\n\nPerfect.\n\nI'm seeing your Hegen screen still here.\n\nWell, let me get back on Hegen. Well, can you see that professional voice clone slide?\n\nYes.\n\nAwesome. Cool. So these are four tracks. Two of these are real recordings from our CEO. Two of them are generated with his AI voice.\n\nWe'll play them all and see if you can guess which are the real ones, which are the AI ones.\n\nWhen we look at our most successful customers, we help them show up and build trust at a scale their own time alone will never allow. I think that is a new era of scaling presence and communication, and it is only possible now.\n\nAnd think about it this way. On average, every person in this world watches seventeen hours of video on YouTube. People don't want to spend those seventeen hours watching AI slob. People want to watch something real, something authentic.\n\nWhen we look at our most successful customers, we help them show up and build trust at a scale their time alone would never allow. I think this is a new era of scaling communication and presence that's only possible in the age of AI.\n\nHowever, when we really watch our customers, how they use the product, how they feel about it, and the problem they are struggling with, that really changed my mind about what focus actually means for a company like ours.\n\nSo can people guess which of these are the real ones versus AI?\n\nYou can comment in the comments.\n\nAnd this is actually a lot of the testing that we do internally as we're focused on, one, making sure people like it, and two, that it sounds very realistic.\n\nWe want this to be a voice that you can use to create videos that your clients, people who know you can watch, and it should sound like you. Because we know a lot of our users are creating videos, they're actually gonna meet with these people in real life. And trust is critical.\n\nSo the realism of the voice so that when they actually meet you in person, it sounds like the same person is critical for us.\n\nAnd honestly, forgot which ones are which.\n\nCan follow-up later with the answers, but yeah, takeaway is this fooled our own people at Hagen very well, who know Josh very well. And goal is that when you use this, you know, your customers, your clients, your your audience will, you know, very much think this is the real you. And, again, it really is because it has two plus hours of your real audio so that it does match how you'd actually deliver lines.\n\nAwesome. With that, I will, open up to q and a.\n\nAwesome. Thanks, Adam.\n\nI love, the reactions of guessing. I think, even if we don't reveal which one is real, it's clear from the chat that it's very hard to tell which one is real.\n\nYeah. What are the guesses?\n\nAll over the place. Some D fake, some D real.\n\nBut, let's start. We've got a couple... We've got we've got lots of questions. Let's start with a basic one, I think probably a lot of people are wondering, if someone already has a voice in Heijen, in their Heijen account, what do they need to do to create it and use the new Heijen voice tools? Can they upgrade it? Do they need to recreate it? What is that process?\n\nYou should be able to use your existing voice with HeyGen today.\n\nWhere you go with that is when you stop edit voice, it'll bring up this menu here.\n\nOr can you see my screen?\n\nWe're still seeing the slide.\n\nLet me reshare. I'll show on the app.\n\nSo tap your existing voice. You can tap edit voice, and that, you know, exists anywhere you're creating videos.\n\nAnd then when you tap edit voice Welcome to you'll be able to switch engines. So let's say you already have an old voice and it's been on eleven labs or, you know, some other engine, you can preview it across every engine and then Hey. Switch it to HeyGen Voice if you want, and you can also fine tune with those settings here.\n\nAnd then you can update these default settings. In this case, my voice is already voice, so it's not letting me update my default. But if you were using an existing voice that's set to some other engine, just open edit voice Hey. Welcome to voice, fine tune the settings till you're happy, and then, you can apply it for that video and and save this default so it uses this voice going forward.\n\nAwesome.\n\nAnd we've got a question too. If someone's using, you know, the Eleven Labs PVC and then they're looking at Hegen, how would you compare Hegen's professional voice clone with the Eleven Labs PVC voice?\n\nGreat question. I would... So we're actually testing with a lot of people using both. We find a lot of people like Heijen's voice for nailing the pronunciations, feeling natural and expressive at the same time.\n\nAnd we're very interested to also get your feedback as well.\n\nSo if you do have an 11labs PVC, one advantage is it's really easy to create your HM PVC. If you already have that thirty minutes, two hours of audio, you can reclone using that exact same audio with HN Voice.\n\nAnd again, if you're interested, we'll offer two months free to people who want to test out a PVC can follow-up with that option.\n\nSo, yeah, I'd recommend try it out yourself and and let us know how you how you think they compare.\n\nAwesome. Thanks, Adam. And, yeah, so for anyone on the call, who's joined us, stayed with us today through the webinar, we will give you access. You can try out the professional voice clone from Hegen for free. So we'd love to know what you all think. We'll follow-up on email with that information.\n\nLots of other good questions in the chat too. We'll do our best to send out resources and would love to see folks at the support office hours that Jane dropped... James dropped the link for in the chat if you do have questions you're looking for more of a walk through. And then, also, we have another webinar later this week on the twenty ninth that'll go into creating looks. It's Halloween themed, but it'll show you the basics of creating multiple different looks, using HeyGen in creative ways. Great to combine with your voice clone. We run these webinars with different pieces so you can build your skills as you go. And we've got intro webinars, again, support office hours, lots of resources.\n\nSo thank you, Adam, for the deep dive today into HeyGen Voice, and we'll share the presentation recording with everyone and access so you can create your professional voice clone and let us know what you think.\n\nThanks, all.\n\nThanks, everyone. See you next time.\n\nBye."}],"hls_enabled":true,"embed_options":{"opaqueControls":null,"playerColor":"#00C3FF","playerColorGradient":{"on":false,"colors":[["#00C3FF",0],["#66ffff",1]]},"plugin":{"captions-v1":{"on":"true"},"chapters":{"on":"true","chapterList":[{"id":"045787a2-3042-448c-a9d5-4387d010b745","time":"153.155","ai_title":"Understanding AI Voice Limitations","title":"Understanding AI Voice Limitations","deleted":"false"},{"id":"1eb019ba-6c8c-4633-b52e-176dc3bdfcaa","time":"74.105","ai_title":"HeyGen Platform Overview","title":"HeyGen Platform Overview","deleted":"false"},{"id":"25a17232-3ead-4f01-af69-8784d148bd70","time":"654.96","ai_title":"Creating Voice Clones in App","title":"Creating Voice Clones in App","deleted":"false"},{"id":"517d93cd-4b00-4852-a0a0-fe6da32b8bf5","time":"262.535","ai_title":"HeyGen Voice Technical Advantages","title":"HeyGen Voice Technical Advantages","deleted":"false"},{"id":"616a8008-d053-4751-a28a-7c80282c34c1","time":"336.49","ai_title":"Video Adaptive Audio Setting","title":"Video Adaptive Audio Setting","deleted":"false"},{"id":"67d9f7ee-87de-4e36-8f62-ff0928886f29","time":"1372.405","ai_title":"Best Practices for Recording","title":"Best Practices for Recording","deleted":"false"},{"id":"690f29ec-8c05-41ca-88b9-103c0d122d14","time":"897.06","ai_title":"Voice Fine Tuning Controls","title":"Voice Fine Tuning Controls","deleted":"false"},{"id":"95777b00-2aa6-4872-8345-7569b6a2e5fd","time":"449.73","ai_title":"Environmental Audio Demo Examples","title":"Environmental Audio Demo Examples","deleted":"false"},{"id":"a021c237-dc37-4bb9-b047-25c7a416ffad","time":"1536.57","ai_title":"Custom Pronunciation and Glossaries","title":"Custom Pronunciation and Glossaries","deleted":"false"},{"id":"dc67330f-c40e-4e18-aee2-c704984e7ee4","time":"2153.82","ai_title":"Q&A and Legacy Voice Upgrades","title":"Q&A and Legacy Voice Upgrades","deleted":"false"},{"id":"deec8c65-c033-4875-86b3-eb8dedc6982f","time":"1821.685","ai_title":"Professional Voice Clone Comparison","title":"Professional Voice Clone Comparison","deleted":"false"},{"id":"f7abd758-99f8-495b-aae2-2e15cd0d5dfc","time":"1094.58","ai_title":"Implementing HeyGen Voice API","title":"Implementing HeyGen Voice API","deleted":"false"}]}},"volumeControl":"true","fullscreenButton":"true","controlsVisibleOnLoad":"true","bpbTime":"false","roundedPlayer":"8","vulcan":true,"stillUrl":"https://embed-ssl.wistia.com/deliveries/67545b31b3a76064cb84ef300fb71154.png","bodyFontFamily":"Inter","contrastIcons":true,"newRoundedIcons":true,"shouldShowCaptionsSettings":true},"embedOptions":{"opaqueControls":null,"playerColor":"#00C3FF","playerColorGradient":{"on":false,"colors":[["#00C3FF",0],["#66ffff",1]]},"plugin":{"captions-v1":{"on":"true"},"chapters":{"on":"true","chapterList":[{"id":"045787a2-3042-448c-a9d5-4387d010b745","time":"153.155","ai_title":"Understanding AI Voice Limitations","title":"Understanding AI Voice Limitations","deleted":"false"},{"id":"1eb019ba-6c8c-4633-b52e-176dc3bdfcaa","time":"74.105","ai_title":"HeyGen Platform Overview","title":"HeyGen Platform Overview","deleted":"false"},{"id":"25a17232-3ead-4f01-af69-8784d148bd70","time":"654.96","ai_title":"Creating Voice Clones in App","title":"Creating Voice Clones in App","deleted":"false"},{"id":"517d93cd-4b00-4852-a0a0-fe6da32b8bf5","time":"262.535","ai_title":"HeyGen Voice Technical Advantages","title":"HeyGen Voice Technical Advantages","deleted":"false"},{"id":"616a8008-d053-4751-a28a-7c80282c34c1","time":"336.49","ai_title":"Video Adaptive Audio Setting","title":"Video Adaptive Audio Setting","deleted":"false"},{"id":"67d9f7ee-87de-4e36-8f62-ff0928886f29","time":"1372.405","ai_title":"Best Practices for Recording","title":"Best Practices for Recording","deleted":"false"},{"id":"690f29ec-8c05-41ca-88b9-103c0d122d14","time":"897.06","ai_title":"Voice Fine Tuning Controls","title":"Voice Fine Tuning Controls","deleted":"false"},{"id":"95777b00-2aa6-4872-8345-7569b6a2e5fd","time":"449.73","ai_title":"Environmental Audio Demo Examples","title":"Environmental Audio Demo Examples","deleted":"false"},{"id":"a021c237-dc37-4bb9-b047-25c7a416ffad","time":"1536.57","ai_title":"Custom Pronunciation and Glossaries","title":"Custom Pronunciation and Glossaries","deleted":"false"},{"id":"dc67330f-c40e-4e18-aee2-c704984e7ee4","time":"2153.82","ai_title":"Q&A and Legacy Voice Upgrades","title":"Q&A and Legacy Voice Upgrades","deleted":"false"},{"id":"deec8c65-c033-4875-86b3-eb8dedc6982f","time":"1821.685","ai_title":"Professional Voice Clone Comparison","title":"Professional Voice Clone Comparison","deleted":"false"},{"id":"f7abd758-99f8-495b-aae2-2e15cd0d5dfc","time":"1094.58","ai_title":"Implementing HeyGen Voice API","title":"Implementing HeyGen Voice API","deleted":"false"}]}},"volumeControl":"true","fullscreenButton":"true","controlsVisibleOnLoad":"true","bpbTime":"false","roundedPlayer":"8","vulcan":true,"stillUrl":"https://embed-ssl.wistia.com/deliveries/67545b31b3a76064cb84ef300fb71154.png","bodyFontFamily":"Inter","contrastIcons":true,"newRoundedIcons":true,"shouldShowCaptionsSettings":true}}}