|
Briefing
|
- AI System to Mimic Voice – Canadian startup Lyrebird developed human voice imitation system using neural network that gathers data and gains speech patterns by analyzing speech recordings
- System Training – Listens to hours of speech audio to comprehend characters, phonemes and word pronunciation correctly, then develops new sentences while adding distinct intonations and emotions
- Comparison to Human Voice – Created voice makes buzzing noise, emits distinct robotic sound, and has no sounds from breathing and mouth movement, making it distinguishable from ordinary human speech
- Potential Negative Implications – Generated audio clip seems genuine when heard by inexperienced ears, potentially creating ethical and safety issues about impersonation
|
