from Hacker News

Show HN: Creates hyper-realistic voice clones from just 3 seconds of audio

by blacktechnology on 1/9/25, 5:39 AM with 4 comments

  • by replwoacause on 1/9/25, 2:48 PM

    I'm not too comfortable giving my voice data over to some random company. Just like I wasn't comfortable giving my DNA over to 23andme (who BTW was hacked, leaking 7 million user's profile data, and is now on the brink of collapse). Am I alone in this kind of thinking? What makes people comfortable using a service like this?
  • by blacktechnology on 1/9/25, 5:43 AM

    Hey everyone! I recently created an AI voice cloning product that can clone any voice with just 3 seconds of original audio! I'd like to share it with you all. Here's how it started: A friend of mine always sends me voice messages, and I wanted to play a little prank on her - what if I could reply using her own voice? With AI technology being so advanced now, I made it happen, haha, and it's been so much fun. Try it out at: https://anyvoice.net/ai-voice-cloning What Makes AnyVoice Special?

    Only needs 3 seconds of audio to clone a voice! Yes, it's that simple. Just record something like "The weather is beautiful today, let's go get dinner" on the webpage - say anything you want. Other solutions out there require dozens of sentences to record, who has time for that... Supports four languages - English, Chinese, Japanese, and Korean: In testing, the results are incredibly natural. When my friend heard her own voice, she was so shocked she thought she was getting a scam call!

    Real Usage Scenarios:

    Prank Your Friends: Reply to their voice messages using their own voice. Just record their voice (most voice messages are longer than 3 seconds, right?), then record it on the website, and it's instantly cloned.

    Pro tip from my experience: Use two phones for easier operation. Phone A plays your friend's voice, Phone B opens https://anyvoice.net/ai-voice-cloning to record and clone the voice. After cloning, Phone B plays the cloned voice while Phone A records a voice message - your friend will be totally shocked!

    Short Video Voiceovers: Content creators know the pain - you either need to find someone or do the voiceover yourself, which is tiring. If you want to use voices from TikTok or other platforms, they often charge fees. We're completely free - you can get any voice you want! Language Learning: Having your own voice read foreign language texts feels amazing and really boosts learning motivation. Pretend to Speak Foreign Languages: Record yourself in English, clone your voice, then input Chinese, Japanese, or Korean text - you'll get your voice speaking these languages!

    How to Use: Super simple interface, just three steps:

    Upload/record a voice sample (3-10 seconds) Choose the text you want to generate, up to 200 characters (it's a free service, please understand - we might increase this limit in a paid version in the future) Wait for generation (usually takes just seconds)

    Try it here: https://anyvoice.net/ai-voice-cloning Finally: We're currently in public beta, completely free with no paid features! Welcome all to try it out and give feedback. If you find it useful, please give it a like Feel free to leave any suggestions or questions in the comments below, I'll respond to each one.