A British firm, Synthesia, is revolutionizing voice cloning with its new AI tool, Express-Voice, which accurately replicates UK regional accents—outperforming many US and Chinese competitors. Traditional AI voice training relies heavily on North American and southern English data, leading to uniform sounds. To address this, Synthesia dedicated a year to compiling a diverse UK voice database from studios and online sources. Customers, including CEOs, are seeking authentic regional representation in training videos and presentations. The challenge remains with less common accents due to limited data. While addressing accent recognition issues in voice-activated AI, concerns arise over speech homogenization and potential misuse by scammers, especially as free, open-source voice-cloning tools proliferate. Though Synthesia plans to implement safeguards against malicious use, the rapid evolution of this technology poses significant challenges for safety and diversity in language representation.
Source link
Share
Read more