How I made voice cloning accessible to everyone
Hey Indie Hackers,
I've been fascinated by voice cloning for quite some time now. It has advanced at a rapid pace over the past several years, but there was one glaring issue with the industry around it... The ease of use for the average person.
If someone wanted to try out voice cloning, they would have to visit some website they've most likely never heard of, figure out how to record a clip of their voice, upload that clip, and finally download an audio file containing the cloned speech. That's a lot of friction for most people.
Several months ago, I thought to myself, "What if voice cloning could be directly integrated into a social platform like Discord?" Excited by the thought, I immediately began working on a prototype for what is now known as CopyKitten.
CopyKitten is a Discord bot that will join your voice call, listen to you speak, and create a clone of your voice. No external website, all integrated directly within Discord. I've tested just about all of the AI voice cloning model providers, and for this project, I decided to go with FishAudio for its great balance of quality, realism, speed, and cost.
I recently released CopyKitten, and things have been going great so far. Plenty of user feedback has made it my way, and I'm always working to further improve the core user experience. If this interests you at all, I would love it if you tested it for yourself and let me know what you think!
Discord integration is smart friction reduction but how do you handle consent? If the bot joins a voice call with 5 people, does everyone need to opt-in or just the person who invited it? [
Before cloning, the bot requests consent from the person and will only join if they click 'agree'. The bot doesn't listen to the audio of every person in the call, only the person it has been asked to clone, otherwise the cloning would be ineffective if a bunch of people were talking at once.
Fun concept! Removing friction by embedding voice cloning directly into Discord is a smart distribution move. Also, I like the black background with colorful elements. The half-white, half-purple cat design is very cool and memorable.
One security question: since the bot joins live voice calls and processes audio, how are you handling consent, storage, and deletion of voice data to prevent misuse or unauthorized cloning of someone’s voice?
With voice tech, trust and abuse prevention are critical.
Thanks! Yeah, those are all valid safety/security concerns.
Consent: When you use the bot to make a cloning request, the bot first requests consent from the user with a message that lays everything out clearly and links to our privacy policy. The user must click 'agree' before the bot is able to join the voice call.
Storage: Raw audio data is not stored on server. It only exists in memory during capture and is then processed into a distilled voice clone.
Deletion: Anyone can delete their voice clone at anytime by using the bot's /delete command in Discord. They can also email me at anytime for deletion. When you make a voice clone, it will only be accessible from within the Discord server on which you create it on.
Hopefully I was able to address each of those sufficiently! Let me know if anything I said raises concern or you feel requires further explanation :D
That’s a thoughtful approach. Explicit in-app consent before joining the call is the right baseline, and not storing raw audio server-side reduces a lot of long-term risk. The scoped access per Discord server is also a good containment decision.
One thing you may want to think about as usage grows is abuse scenarios inside shared servers. For example, rate limits per user, per server, and monitoring for rapid clone creation patterns can help reduce impersonation risks. Also worth logging clone creation and deletion events with basic audit trails so you can investigate disputes if they happen.
Voice tech moves fast, but trust moves slower. The more transparent and restrictive the defaults, the stronger your positioning.
We’re a security team building Nautillo Pro to help founders test real attack paths before products scale. If you ever want an external security view of web components, you’re welcome to try our web attack simulator. There’s a free version for solo founders, so you can run checks regularly as you ship new features.
Cool concept overall! Distribution through Discord is a smart angle.
Congrats on the launch, looks solid. How are you currently thinking about acquiring early users and gathering feedback?
Thank you! So far, I've grown the bot to 3000 servers by marketing it on Reddit, YouTube, Instagram, TikTok, X, HackerNews, ProductHunt, Discord bot directories, and SEO articles. Feedback has come mostly through Discord and email.
3,000 servers is impressive, congrats, that’s real traction 👏
Since Reddit is already working for you, I’m curious, are you mostly posting launch-style promos, or are you also embedding the bot into organic discussions inside niche subreddits? I’ve found that for Discord tools especially, the highest conversions usually come from problem-thread positioning rather than announcement posts.
If you’d like, I can take a quick look at where you’ve been posting and suggest a few subreddit + angle tweaks that could increase signups without increasing volume.
Great concept and clean execution 👏
This has clear potential in the market.
We support apps, tools, and web platforms with real creator-led growth and user acquisition 🚀
Happy to connect.