Parakeet Inc. Releases "Paramatch" Speaker Identification Model Demo to Identify Original Voice Actors from Unauthorized AI-Learned Synthetic Voices

Parakeet Inc. (Headquarters: Hirono-machi, Futaba-gun, Fukushima; Representative Director: Taiki Nakamura) has released a demo version of its speaker identification model, "Paramatch," on Hugging Face for free. This model is a speaker identification AI that predicts and analyzes the top 3 closest matching speakers when an audio file is uploaded. It was developed to provide clues for identifying the original voice actor whose voice was used to create AI-generated synthetic voices from unauthorized AI voice synthesis models. The demo URL is https://huggingface.co/spaces/Parakeet-Inc/paramatch. The company states that while AI learning for information analysis is broadly permitted under Article 30-4 of Japan's Copyright Act, distributing/publishing unauthorized AI voice synthesis models, offering third parties the ability to freely generate speech in a specific voice actor's voice, or publicly distributing/disseminating synthetic voices generated in that voice actor's voice constitutes a significant infringement on the identity and rights of performers. Paramatch predicts the top 3 speakers with the closest voice quality from its database and provides similarity scores. Key features include predicting similar speakers from a database and supporting audio files under 30 seconds. Uploaded audio is not saved on the server, nor is it used for secondary purposes. Learning data is solely used for building the Paramatch model and not transferred to other generative models or projects. The company stresses that the tool's results are for reference only and not a basis for legal judgment, and continuous improvement is ongoing. The demo aims to validate the effectiveness of the technology as a countermeasure for protecting voice rights and to initiate dialogue with voice actors, agencies, and rights organizations. Parakeet Inc. has a track record of establishing fair revenue sharing frameworks with IP holders and voice actor agencies through legitimate licensing agreements for its real-time AI voice changer "Paravo." The company seeks collaboration to validate effectiveness against real unauthorized uses, improve database accuracy, provide technical advice for industry guidelines, and participate in joint projects for performer rights protection. Company Name: Parakeet Inc. Representative: Taiki Nakamura Location: Hirono-machi, Futaba-gun, Fukushima Business: Research and development of speech synthesis and speech conversion. Main Products: Real-time voice conversion engine "Paravo," Text-to-speech engine "Paratts," Speaker identification model "Paramatch," Voice dubbing engine "Patra," Furigana prediction engine "Parayomi." Company Website: https://parakeet-inc.com/.