A developer and AI lab going by abus-aikorea just open sourced one of the most complete voice AI toolkits released in 2026, and the project is the first credible all in one replacement for the paid voice stacks that creators have been paying subscriptions for over the last two years.
The repository is called Voice-Pro, and it consolidates zero shot voice cloning, Whisper based transcription, YouTube downloading, vocal isolation, and multilingual dubbing into a single self hosted application. The interface is a Gradio WebUI, the whole thing runs locally, and it supports more than 100 languages for dubbing and translation. The entire project is 100 percent open source.
This is not a single feature tool. It is the entire ElevenLabs plus Descript workflow rebuilt as an open source stack, and the value of having all of that in one place without a subscription is hard to overstate.
The Problem Most Voice AI Tools Leave on the Floor
The current state of voice AI for creators in 2026 is fragmented and expensive. The workflows that creators actually need are split across multiple paid services, each with its own subscription, its own UI, and its own data policy.
- Voice cloning lives behind ElevenLabs or Resemble AI subscriptions that charge per minute or per character
- Transcription requires a separate Whisper deployment or a paid service like Otter or Rev
- Vocal isolation needs another tool like Lalal.ai or a dedicated DAW plugin
- Dubbing and translation are split across services like Rask, HeyGen, or a manual Descript pipeline
- YouTube downloading requires a separate utility just to bring source media into the workflow
- Privacy and data ownership are unclear when source audio is uploaded to multiple third parties
Voice-Pro is built to consolidate all of these into a single self hosted application, and that consolidation is what makes the project a meaningful contribution to the creator tooling ecosystem.
What is Actually in the Repository
The repo ships a complete voice AI platform with a clear, focused feature set rather than a single trick demo.
- Zero Shot Voice Cloning: Voice-Pro can clone a voice from a short reference sample without fine tuning, which mirrors the core capability of ElevenLabs and similar services
- Whisper Transcription: The transcription layer uses OpenAI’s Whisper model for high quality speech to text, with multilingual support
- Vocal Isolation: The tool includes source separation so vocals can be isolated from background music and noise, which is the Descript and Lalal.ai feature set rebuilt locally
- YouTube Downloading: Built in YouTube downloading means creators can bring source media into the workflow without leaving the app or using a separate downloader
- Multilingual Dubbing: The dubbing pipeline supports 100 plus languages, which covers the use case of translating and re voicing content for international audiences
- Gradio WebUI: The whole thing runs through a Gradio based web interface, which makes it approachable for non technical creators while still being fully self hosted
This is the rare project that does not try to be the best at one thing. It tries to be a complete stack, and the integration between the features is where the value shows up.
Who This is for
The audience for Voice-Pro is broad, because voice work is broad, and the project covers the workflows that almost every creator and educator runs into.
- Content Creators: YouTubers, TikTokers, and short form video creators who need voiceovers, dubs, and clean transcripts without paying multiple subscriptions
- Podcasters: Podcast hosts who want local transcription, vocal cleanup, and editing in one tool rather than stitching together three paid services
- Video Editors: Editors producing content for clients who need a self hosted workflow for sensitive audio that cannot be uploaded to third party AI services
- Translators: Professional translators and localization teams who need dubbing and translation tooling they can run on their own infrastructure
- Educators: Teachers, course creators, and educational content producers who want multilingual dubs of their lectures and lessons
- AI Developers: Engineers building voice AI features can use Voice-Pro as a reference implementation of a full voice pipeline they can fork and customize
If you have ever paid for an ElevenLabs subscription and a Descript subscription and a Lalal.ai subscription in the same month, Voice-Pro is built for you.
What People are Saying
The community reaction has been quick and appreciative, which is the right response to a project that fills a real gap.
“Nice find”
@travelsoultan
The reaction captures the sentiment well. Voice-Pro is the kind of repository that creators stumble across and immediately recognize as a serious contribution. The combination of features is the selling point, and the open source license is what makes the find worth sharing.
Why This Repo Matters for the Creator Tooling Ecosystem
The bigger story is not any single feature. It is what becomes possible when the entire voice AI stack is free and self hosted.
- Creator Cost Drops to Zero: The monthly bill for ElevenLabs, Descript, and a vocal isolation tool can easily reach $50 to $200 per creator. Voice-Pro replaces all of that with a free self hosted option
- Privacy Becomes the Default: Running locally means voice recordings, transcripts, and dubbed content never leave the creator’s machine, which matters for journalists, educators, and anyone working with sensitive audio
- Multilingual Production Becomes Accessible: Dubbing in 100 plus languages at zero marginal cost means a creator in any country can publish for any audience, which levels the global creator playing field
- Open Source Beats Vendor Lock In: The toolkit is open source, so the creator community owns the workflow rather than depending on a paid vendor’s roadmap or pricing changes
- The Workflow Becomes Composable: Because the whole stack is open, builders can integrate Voice-Pro into their own apps, scripts, or content pipelines without negotiating API access
Repo: https://github.com/abus-aikorea/voice-pro
Voice-Pro is the open source release that turns the phrase “AI voice stack” from a paid service category into a self hosted toolkit, and that distinction matters more than it sounds. The project is not a polished commercial product, and it is not trying to be, but it is the most complete open source voice AI platform available in 2026, and it covers the workflows that creators and educators actually need.
Anyone producing audio or video content should clone the repo, run it locally, and try the workflow on a real project before paying another month of subscription fees to three separate services. The era of paying $200 a month to stitch together four AI tools is ending, and Voice-Pro is the open source proof. Run it locally, own your audio, and keep your subscription budget for something else.
