Microsoft’s new AI can simulate anyone’s voice with 3 seconds of audio

Summary

KEY POINTS:

1. VALL-E replicates emotional tone and acoustic settings.
2. Applications encompass text-to-speech and speech editing.
3. It generates discrete audio codec codes from prompts.
4. Microsoft hasn't released VALL-E's code due to potential misuse.
5. Researchers propose a detection model and commit to Microsoft AI Principles.

See: VALL-E (X) (microsoft.com)

References

• Media Check Ars Technica - Media Bias/Fact Check (HIGH CREDIBILITY)
• Collections | The Runway (airforce.gov.au)
• ADDITIONAL READING RAAF RUNWAY (PME)
• RAAF RUNWAY: RATIONALE, GUIDELINES, LEARNING OUTCOMES, ETC