76

Well, Apple did a good job with Speech synth Alex. Wonder who could be behind this. Is it based off human voices, or Apple written it completely from scratch? We never know what Apple did with this synth.

Thumbs up

77

I think it is safe to say that Apple did use someone to concatenate his voice. Obviously he recorded a whole bunch of words, phrases, syllables, breath sounds, etc, sort of like Vocaloid. But to make it really high quality, that is another thing. Maybe Apple has some secret algorithm. But the fact that Apple is secretive about many things is the reason why they are one of the top leading money-making companies in the world.

Ulysses
AKA Green_Gables_fan and HeavenlyHarmony
My new, self-hosted version of WordPress!

Thumbs up

78

Microsoft Sam is diphone based, and eloquence is formant based. They don't use any combination of both.
There is a type of synthesis where you record some speech and then the recordings are used to create an Approximation of there vocal tract.
This type of synthesis is called Hidden Markov Models and is used in engines such as modern Microsoft and Google TTS.

Thumbs up

79

The question is: how? Can you give us an example of how you would take someone's recording and make a vocal tract out of it? What software programmes would you need to do it? Could you do that with Audacity?

Ulysses
AKA Green_Gables_fan and HeavenlyHarmony
My new, self-hosted version of WordPress!

Thumbs up

80

and does any one know who voiced amazon alexa?

watch my brother's twitch stream here:
https://www.twitch.tv/sylvrexe

Thumbs up

81

this is exactly the question I also have. recording hours and hours of audio, is not the problem,
and, i am willing to do this. as, to my understanding,
the more  phrases,
there is in a recorded audio, the better for the enjin, but the question remains, what app can we use that can put this all together, so that we can install and test it. if I can find that, I would be perfectly willing to record as many hours of audio as it needs.
I'd even be willing to dissect, cut up,  and do what ever I need to do, with this audio. to make the enjen as good as it could be. the idea that I get, from reading up, is that, the more and more recordings you have, the easier it is for the enjen to provide a natural voice, at the end of the day. so, if only we new what to use, to make this work, I'd be super happy to give this a go.

There's a place for me in this universe. and our journey continues on, together

Thumbs up

82

@75. ESpeak, from what I've read, is a mix of formant and recorded speech. The consonants are gathered from recorded speech samples, whereas the vowels are done by adding together sine waves to produce the harmonics for them.

It's a toaster!

Thumbs up

83

Hmm, interesting. I'd like to see how I can generate an eSpeak sound using Audacity, if I can figure out the amplitude of each harmonic overtone, and how many of them there are.

Ulysses
AKA Green_Gables_fan and HeavenlyHarmony
My new, self-hosted version of WordPress!

Thumbs up

84

ammericandad2005 wrote:

and does any one know who voiced amazon alexa?

Alexa is a computer generated voice. It is not from recordings of any real person(s) voice. It is generated from rules converting text in human speech.

Pics or it didn’t happen

Thumbs up

85

How did you find out that Alexa is purely formant? It surely doesn't sound formant to me at all. I wonder what they used to create it?

Ulysses
AKA Green_Gables_fan and HeavenlyHarmony
My new, self-hosted version of WordPress!

Thumbs up

86

Green Gables Fan wrote:

How did you find out that Alexa is purely formant? It surely doesn't sound formant to me at all. I wonder what they used to create it?

https://developer.amazon.com/docs/custo … rence.html

Pics or it didn’t happen

Thumbs up

87

Here's a video with the voice behind Google.

https://www.youtube.com/watch?v=qnGNfz7JiZ8

Kingdom of Loathing name JB77

Thumbs up

88

the voice of the bbc micro's speech synthesis card was provided by retired BBC News Anker kennith kendel.

watch my brother's twitch stream here:
https://www.twitch.tv/sylvrexe

Thumbs up

89 (edited by akse0435 2018-04-27 20:57:04)

Hi,

Well, I don't know, if anyone is interested, but I found some of the voices behind some danish TTS voices. The links with the voice names links to a danish site called "Speaker- og indlæserforeningen".

Acapela Mette:
Karen Abrahamsen

RealSpeak / Vocalizer Ida:
Anne-Mette Johansen

Vocalizer Expressive Sara:
Jette Sivertsen

Loquendo Frida:
Dorthe Mikkelsen Hansen

Ivona Mads:
Lars Thiesgaard

Best regards> Aksel Christoffersen

Thumbs up

90

Dennis klat (of dectalk fame) was the voice of steven hawking's trusty prose2000 speech generating device.

watch my brother's twitch stream here:
https://www.twitch.tv/sylvrexe

Thumbs up

91

I wonder how Dennis Klatt's real voice sounded, and if it sounded similar to Dectalk. Maybe it could be the voice, that announce the recordings in the Klatt's history of speech history on the website linked in post 75, but we can't know for sure. Does anyone have a recording of Klatt's real voice or something?

Thumbs up

92

Acapela's Josh is voiced by Emerson.
You can hear him in this video.
https://www.youtube.com/watch?v=hWwjZuarB3o
Here is his youtube channel.
https://www.youtube.com/user/11e22m

Kingdom of Loathing name JB77

Thumbs up +1

93

How can i play this file:
https://www.dropbox.com/s/qeqx3h46sblcamv/dbpcmbec?dl=1

Thumbs up

94 (edited by akse0435 2018-10-09 13:35:02)

What is this file?

To me the name indicated, that it could possibly be a database of PCM sound for some speech synthesizer. Is this corect?

Thumbs up

95 (edited by Hayri Tulumcu 2018-10-13 18:36:18)

yes. It is from AT&T natural voices

Thumbs up