VALL-E
{{Short description|Speech synthesis software}}
{{Use American English|date=May 2023}}
{{Use mdy dates|date=July 2023}}
{{Infobox software
| name = VALL-E
| screenshot =
| logo =
| developer = Microsoft
| released =
| latest release version =
| programming language =
| engine =
| platform = Cloud computing platforms
| genre =
| website= https://www.microsoft.com/en-us/research/project/vall-e-x/
}}
{{Machine learning}}
VALL-E is a generative artificial intelligence system for speech synthesis developed by Microsoft Research and announced on January 5, 2023.{{Cite web |title=Microsoft Unveils VALL-E, a Game-Changing TTS Language Model |url=https://www.infoq.com/news/2023/01/microsoft-text-to-speech-valle/ |access-date=2023-09-19 |website=InfoQ |language=en|first=Daniel|last=Dominguez|date=January 27, 2023}} It can "recreate any voice from a three-second sample clip".{{Cite web |last=Morrison |first=Ryan |date=2023-01-10 |title=Microsoft's new VALL-E AI can clone your voice from a three-second audio clip |url=https://techmonitor.ai/technology/ai-and-automation/vall-e-synthetic-voice-ai-microsoft |access-date=2023-09-19 |website=Tech Monitor |language=en-US}} It has been trained on 60,000 hours of English language speech from Meta’s audio library LibriLight.{{Cite web |last=Wodecki |first=Ben |date=January 11, 2023 |title=Microsoft's VALL-E Generates Speech From Just 3 Seconds of Audio |url=https://aibusiness.com/microsoft/microsoft-s-vall-e-generates-speech-from-just-3-seconds-of-audio|website=AI Business}}
See also
External links
- {{Official|https://www.microsoft.com/en-us/research/project/vall-e-x/}}
References
{{reflist}}
{{Compu-ai-stub}}
{{Microsoft-software-stub}}
{{Speech synthesis}}