In recent months, various technology companies have launched different artificial intelligence systems, and, as expected, the great Microsoft He was not going to be left behind, which is why he presented VALLEYAI that has the ability to copy your voice almost identically after only listening to you speak for three seconds.
Recently, Microsoft has developed its approach to modeling Language for Text to Speech Synthesis (TTS). In this sense, the developers have stated that this artificial intelligence called VALL-E can even imitate emotions, acoustic environment and different emphasis in sentences.
“Specifically, we trained a neural codec language model using discrete codes derived from an out-of-the-box neural audio codec model, and considered TTS as a conditional language modeling task rather than continuous signal regression as in work. former. During the pre-training stage, we scaled the TTS training data to 60,000 hours of speaking in English, which is hundreds of times larger than existing systems,” detailed the technology giant founded by Bill Gates and Paul Allen.
In addition, Microsoft is working with VALL-E so that it can work with other generative artificial intelligence models, such as GPT-3 (autoregressive language model that uses deep learning to produce texts that simulate the writing of humans).
In this context, it will be necessary to bring up the fact that the technology company recently announced that it will enable ChatGPT within its main solutions. In particular, the corporate led by Satya Nadella indicated that it will reach Bing during the first three months of this 2023.
Likewise, the company highlights that “VALL-E could preserve the emotion of the speaker and the acoustic environment of the acoustic message in synthesis”.
Due to the above, Microsoft has affirmed that its new artificial intelligence significantly exceeds the latest generation zero-shot TTS system, this in what has to do with the naturalness of speech and similarity of the person the AI imitates.
We recommend you read:
It should be noted that you can check the VALL-E in the Microsoft web portalwhere the original sample of the voice and imitations are included, as well as reproductions of it changing tone or emotions.
#Microsofts #VALLE #Artificial #Intelligence #imitates #voice