Google unveils a terrifying AI tool capable of real-time voice replication.
Google introduced SoundStorm – a model that is cutting edge for non-autoregressive and efficient audio generation.
It uses bidirectional attention, confidence-based parallel decoding and parallel processing to produce high-quality audio. This reduces the generation time by a significant amount.
It can also synthesize dialogues.
Source:
SoundStorm: Google Unveils Terrifying AI Tool Capable of Real-Time Voice Replication
