AnyVoice

AnyVoice is a web-based AI voice cloning and text-to-speech service that creates a voice clone from as little as 3 seconds of audio. A user uploads an audio file or records a sample in the browser, and the generated voice can then read typed text. It offers text to speech, voice cloning and voice design, and is aimed at content creators who need voiceovers without lengthy recordings.

AnyVoice recommends a clear, single-speaker recording of 3-10 seconds at a normal pace with little background noise; a standard smartphone recording is enough. The models support English, Chinese (Mandarin), Japanese and Korean and are trained for natural pronunciation and intonation. Audio is generated instantly and can be downloaded as MP3 or WAV. A voice model library lets users browse voice models shared by the community, and generation history and personal voice models are kept in the account.

The Free plan costs $0 and allows up to 120 characters per generation, 900 seconds (15 minutes) of audio per 30-day cycle, up to 5 voice clone models and 5 voice designs, with slower generation and personal, non-commercial use only. The Pro plan at $14.99/month raises the limit to 1000 characters per generation, allows commercial use, unlimited generations and unlimited voice clone models, with priority generation and voice design access; yearly billing is 33% off.

Voice style customisation is not currently supported, and an API is not yet available, though programmatic access is planned. AnyVoice prohibits use for impersonation, fraud, hate speech or spam and requires consent before cloning another person's voice. Audio data is handled securely within the service. Support is by email, with replies aimed at within one business day.

Use cases

• Voiceovers for content creation without long recording sessions
• Rapid prototyping, dynamic content and real-time applications
• Commercial voice projects on the Pro plan