Parakeet Inc. (a University of Tokyo AI startup) released v3.0 of its real-time AI voice changer Paravo on June 30, 2026. Running on CPU alone with latency as low as 0.06 seconds, the tool can transform your voice into over 100 different voices and is already used by more than 80,000 people.

CPU-Powered Real-Time Conversion in Just 0.06 Seconds

Traditional AI voice changers required expensive GPU computing resources and often introduced over one second of latency for natural-sounding output. Paravo was designed from the ground up to solve this problem with a proprietary AI model optimized for ultra-low-latency real-time conversion on CPUs.

Version 3.0 introduces multilingual support and a redesigned GUI, enabling seamless voice changing during gaming and calls on any device, including laptops. With no practice required and an intuitive interface, the tool has already attracted over 80,000 users.

100+ Voices Including Zundamon and Nekomoto Pato

Paravo features popular synthetic voice characters like “Zundamon” and “Tsukuyomi-chan,” as well as collaboration plans with VTuber “Nekomoto Pato” and more. Users can convert their voice into those of over 100 professional voice actors. A blend feature lets users select up to three voices and adjust their ratios to create a custom voice.

Five characters with 14 voice styles, including “Zundamon,” are available for free as part of the base features. Additional voices can be purchased per character, and since AI models are updated approximately every two to three months, conversion quality continues to improve after purchase.

All Voice Data Officially Licensed, AILAS Member

Parakeet Inc. is a corporate member of AILAS (AI Learning Audio Service), a general incorporated association. Under the principle of “fair trade between people and voice AI,” the company has formal contracts with all voice rights holders registered on Paravo and shares revenue based on usage.

As concerns about rights management grow alongside advances in AI voice technology, Paravo stands out as a voice changer that users can trust, built on legitimate licensing agreements. We previously covered NTT TechnoCross’s AI voice conversion software “SynoraVoice”, and the range of AI voice conversion options continues to expand.

Smartphone Version Planned This Year, Character Voice Submissions Welcome

A smartphone version is planned for release within the year. The company reports that it has already achieved PC-equivalent conversion performance on mobile devices, which will bring real-time voice changing to more scenarios including on-the-go streaming and live events.

Parakeet is also accepting character voice submissions from both companies and individuals. The company develops AI models from submitted voice data and returns a portion of revenue to rights holders. Those interested in having their voice or character’s voice available as a real-time voice changer can find more details on the Parakeet official website.

Commentary

Achieving real-time voice conversion without a GPU is a significant technical advancement that dramatically lowers the barrier to entry for streamers and VTubers. With 80,000 users already on board, plus transparent rights management through official licensing and AILAS membership, Paravo has established strong credibility. If the smartphone version delivers on its promise, it could rapidly expand into mobile streaming and voice chat applications. As competition in the AI voice conversion space intensifies, Paravo’s continued evolution is worth following closely.

Source: PR TIMES