How is Siri's machine voice made and making conversation possible?

ByYogesh Mhatre

Secretarial function application software for iOSSiriIs an utterance analysis / recognition interface that will activate the iPhone's features just by talking to you. It is summarized how Siri 's speech capable of such simple conversation was born.

Machine language: how Siri found its voice | The Verge
http://www.theverge.com/2013/9/17/4596374/machine-language-how-siri-found-its-voice

One day in 2011, MTV, Domino PizzaNickelodeonAs a 37-year-old actress with experience of appearing in a famous CM, such as Dei, for workNUANCEI visited the company. she is"Text-to-Speech(Text-to-speech system) "Although I was only informed about the work of the product, she was able to give a voice in the early twenties to early twenties in the previous CM, so also with regard to the work of Text-to-Speech It was expected.


And she is a small soundproof room that reads "Alice in Wonderland" one by one as long as she can sit for 8 to 6 hours a day,AP WireI randomly read the sentences from the news on the news site (news site) and continued reading the same word as "One! One.One?" With different accents.

She did not expect that the work to make machine sounds is so severe, the voice has completely withered on the 4th day, so we have to take a break, the efforts of the voice actor who is not as rare Seems to be the basis of Siri's voice.


Google purchased a British speech synthesis production company and Amazon for Kindle FireIvonaEach company has made a huge investment, such as buying a company, the system developed originally for disabled people now is a general system that can go to Starbucks just by speaking to the device and car navigation .


Mr. Brant Ward, senior manager of the latest voice design and development division of NUANCE, independent of speech recognition and the Text-to-Speech industry, became the world's largest company, said, "I do not think that he is talking to the computer Please say "Please.

As words have different meanings depending on their strengths, the voice of a robot like a stick reading seen in movies etc. is hard to hear and I feel cognitive dissonance. According to the fact that it makes me think that he is talking to the computer, it may not be recognized that the change appears in the user's way of talking.


It was a difficult task to synthesize machine sounds so that they sound naturally. Simply joining the dictionary words does not result in a practical synthesized speech,PittsburghMake Text-to-Speech ofCepstralAdam Weiment, a company engineer, said: "Vibrations pass through the vocal cords, leak out of the nose, echo around the tongue and sound like a voice.

Voice itself is beautifulRectangular waveIt is important not to vibrate with a soft tissue, "he says. Synthesis by synthesizer can not synthesize human speech, even children,Speak & amp; SpellI do not believe that the toy in fact is actually thinking and speaking.


Computers are increasingly leaping in the technology of synthesized speech, instantly obtaining the speed at which the right combination of words can be searched from a huge database to every corner. originally"2001 Space JourneyTalking computerHAL 9000From the inspiration from the vocoder was used.

Siri's synthesized speech requires a human voice such as a voice actor. Mr. Ward says, "By synthesizing a little synthesizer speech to human voice, it will be a natural synthesized speech, but now we will be able to synthesize natural synthesized speech by synthesizer alone." Also, the goal of Mr. Benjamin Manson, professor of speech and hearing science at the University of Minnesota, is to provide Siri with emotional and social characteristics and insight, important for normal speech, which is called "para linguistics" .


Some actresses who also work as narrators like Mr. Dey and Alison Dafti, but few people publicly talk about the work they received for machine language.

Voice Actor Talent CompanyVoices.com"In our industry, Text-to-Speech is considered a threat and I think that it is still lousy, but the development of machine languages ​​is a human voice actor We may replace it, "he says about the possibility that the voice industry will be atrophied.