FPT’s new AI voice, can you tell the difference between machine voice and human voice?

21

FPT.AI Speech, the technology of converting Vietnamese text into speech (Text-to-speech) of FPT can turn emotionless words into a voice like real people. From January 2019, FPT.AI Speech updated 3 new Vietnamese voice versions for natural voice quality like the human.

Specifically, three updated voice versions are Northern male voice (Le Minh), Northern female (Ban Mai) and Southern female voice (Lan Nhi). New voices for higher quality, with natural readings like humans, can automatically be breaking off when reading long sentences, combined with expressiveness, are completely different from regular, unnatural, regular speech machine languages.

Please click HERE to listen to FPT’s audition.

To produce the most “real” voice versions, the experts and engineers of FPT Technology Innovation Department (FTI) have been studying for a long time, collecting, analyzing and processing millions of data and thousands of male and female voices from the North, Central and South regions. At the same time, they develop and apply Linguistics, Acoustics and technologies: Natural Language Processing, Machine Learning, and Digital Signal Processing.

Customers can demo technology directly on FPT.AI Speech website.

FPT.AI Speech is developed exclusively for Vietnamese people, based on in-depth research on specific voices of each region in Vietnam. Previously, FPT.AI Speech had 4 different voice versions, including Northern female voice, Northern male voice, Hue female voice and Southern male voice. In particular, all voice versions are free for everyone to access and use, but if customers need to handle a large amount of data, FPT will offer cost packages to customers so that they can get the best technical support.

Practical application

FPT.AI Text-To-Speech brings many practical values and applications in customer care switchboard, call center, or radio systems in public areas such as transportation, health care… Applying Amazon cloud technology, API of FPT.AI can quickly respond to all customer requests, meeting the demand of sudden loading of the system. With the advantages of natural, human-like, good quality voice, most of the Vietnamese voices on youtube, some customer support switchboards or reading applications in Vietnam… are using use FPT’s voice.

Around the world, Text-To-Speech is a speech processing technology that has been studied for several decades, especially developed in the last 10-15 years. This technology has grown almost completely overseas, especially in English-speaking countries. When applied to life, this technology brings great benefits to the disabled community, such as the visually impaired, who lose the ability to read, speak or express.

In Vietnam, Text-To-Speech is taking the first steps, FPT.AI Speech is endeavored to research and give birth by FPT Technology Innovation Department (FTI) to be present in Vietnamese lives.

You can connect to FPT.AI’s API or convert text directly on the website’s user interface into audio files under popular formats like MP3, WAV,…

So, with the updated voice version, FPT.AI Speech promises to bring the product to businesses in general and individuals who need to use good and customer friendly technology to convert Vietnamese text into Vietnamese speech to optimize production and business.

Currently, FPT Technology Innovation owns the platform of FPT.AI artificial intelligence, applied to the research and development of FPT.AI Speech – technology to convert text into speech and voice recognition. FPT.AI Text-to-speech is researched by a team of experienced scientists, experts, doctors, engineers graduated from prestigious Universities such as Cambridge University, University of Versailles, National University Pusan, Stuttgart University, and Nanyang Technological University… and have been trusted by many businesses to optimize costs and operations.

Phuong Thao

Related posts: