 |
| Microsoft Restricts Release of Advanced VALL-E 2 AI Voice Synthesizer Over Misuse Concerns |
In a move reminiscent of OpenAI's 2019 decision to withhold full research on GPT-2 due to its potential dangers, Microsoft announced on Thursday that its new VALL-E 2 voice synthesizer AI will not be publicly released. The VALL-E 2 system, a zero-shot text-to-speech synthesis (TTS) AI, can recreate hyper-realistic speech from just a few seconds of sample audio, surpassing previous systems in speech robustness, naturalness, and speaker similarity.
The VALL-E 2 AI can even handle complex sentences and tongue twisters, making it a groundbreaking development in the field. Potential beneficial uses for the system include enabling speech for individuals with conditions like aphasia or Amyotrophic lateral sclerosis (ALS), as well as applications in education, entertainment, journalism, chatbots, translation, accessibility features, and interactive voice response systems like Siri. However, the team acknowledges the significant risk of misuse, such as spoofing voice identification or impersonating specific speakers.
Due to these risks, Microsoft has decided to limit VALL-E 2's availability to research purposes only. "Currently, we have no plans to incorporate VALL-E 2 into a product or expand access to the public," the team stated. They have also provided a Report Abuse Portal for anyone who suspects misuse of the AI.
Microsoft is not alone in this endeavor. Other companies like Google with its Chirp, ElevenLabs with Iconic Voices, and Meta with Voicebox are also developing similar voice synthesis technologies. However, these systems have faced ethical scrutiny, particularly because they have been used to scam individuals by emulating the voices of loved ones or celebrities. Unlike AI-generated images, there is currently no effective way to watermark AI-generated audio, exacerbating concerns about misuse.
0 Comments