[Not LV] VocaliD: Human Voicebank Initiative

praChirp · Post by **praChirp** » May 6th, 2014, 9:44 am

Has anyone heard of this before? Did you consider actually it? Has anyone actually donated?

The url is here: http://vocalid.org/
And there's also a ted talk: http://www.npr.org/2014/03/07/283452215/how-do-you-construct-a-voice

-Prachi

Cori · Post by **Cori** » May 6th, 2014, 10:02 am

I heard of it a couple of months ago. I briefly considered it, but didn't take it further. They're quite offputting, I thought, in how they're asking for help (especially given they go on to say "but we're looking at recording through an iPhone app" which wouldn't usually spring to my mind as being 'very consistent high quality' audio.)

Q: I have some speech I recorded for another purpose. Can I just send that to you to build a synthetic voice?

A: Typically that won’t work very well. First, unless you recorded many hours of speech, the recordings may not have enough examples of all the different speech sounds and sound sequences needed to make a synthetic voice. Even if you do have many hours of recorded speech, unless you have very consistent high-quality audio with accurate transcriptions of what you recorded, it may be prohibitively difficult to use the recordings.

I do, I do, and I do ... so, perhaps it won't be prohibitively difficult.

But if I'm to put 2-3 hours into something, I have to admit, I'd rather it was a book.

Post by **chocoholic** » June 3rd, 2014, 6:44 am

I filled out their contact form a few weeks ago but never heard back. I am pretty sure the form indicated that they weren't quite ready for volunteers but that you could leave your name if you were interested. I had a few suggestions for them, one of which was that they might want to post here (or have me do it for them) when they're ready for readers, since we have loads of people who already have audio recording equipment and are used to recording for free. I also suggested that if they could reduce the total time required, that would be helpful to most prospective readers -- it wasn't clear to me whether "2-3 hrs" meant 2-3 hrs total or 2-3 hrs finished audio, which as we all know could take a lot longer.

I mentioned that LV has thousands of hours of PD audio already available, so they have at least heard about us, if anyone there is reading the comments on the contact forms.

I don't know if the language in our older books would be useful to them or if it's just easier to build an algorithm around a prescribed set of words that are all the same for each reader.

MClavie · Post by **MClavie** » June 24th, 2014, 12:05 pm

I'm very interested in recording for them and I, too, filled out their contact form a few weeks ago. I haven't heard back either, but their website said that they were still working on getting set up for remote recording - so far they've only collected the voice donations in-person at their lab. I'm hoping they'll be able to set up that infrastructure soon and will get back to those of us who filled out the contact form at that time.

I got the impression that they would want specific recordings and that their algorithm would work around that particular set of words, but I don't know that for sure. I also thought that they would want raw recordings and would do their own editing as needed (so the 2-3 hours would be total recording time), but that was just my assumption.

Here's hoping they're up and running soon. It seems like there are plenty of people waiting who would love to receive a customized voice.

RuthieG · Post by **RuthieG** » June 24th, 2014, 12:48 pm

It looks as if the demon money is the sticking point:

Their registration email wrote: Thank you for wanting to be part of VocaliD and The Human Voicebank
Initiative - our effort to collect voice samples from donors like
yourself.

At present, we are raising funds to launch the voice drive and store
all the data. We are also developing tools to collect high quality
speech recordings without requiring donors to visit a professional
studio. We hope to have the tools ready in the near future and will
keep you abreast on our progress. In the mean time, we could use your
help in spreading the word about the project and recruiting others who
may want to contribute.

Together we can change lives!

If it's an ithingy app, that's me out

.

Ruth

MClavie · Post by **MClavie** » May 1st, 2015, 9:54 am

Hi everyone,
I just received an email from VocalID; apparently they have been continuing to work on this initiative and are now ready to collect voice donations. They are aiming to bring their voices to market in 2016. You can check it out and record your voice donation at www.vocalid.co/voicebank.

I'm so glad to see this finally going somewhere; I think it will help a lot of people. Apparently they have already gotten a lot of feedback from both potential donors and recipients of voices.

~Megan

tovarisch · Post by **tovarisch** » May 1st, 2015, 10:48 am

On that page they wrote:Will my recordings and information be shared with others?

We understand your personal data is sensitive information, and we make confidentiality a priority. To store and retrieve recordings and to match donor-recipient voices, we gather information such as name, age, gender, height, primary language, other language, email, etc.

And the answer is ... ? Are they just ignoring the question? A priority, yes, I get that. They collect the information, yes, obviously. Are they going to share it??? ...

I am not a native English speaker, so I won't participate for now. And when they are ready to expand into Russian, I really hope that the question of how they are going to treat my info and my voice is going to be answered, and not avoided. Just something to ponder...

MClavie · Post by **MClavie** » May 1st, 2015, 2:04 pm

That's a really good question, Tovarisch. There is a "contact us" button that leads to an email address; one could ask that question before submitting a recording.

In case anyone is interested, here is the full text of the email that I received from VocalID:

Thank you for your interest in VocaliD. We haven’t forgotten about you.
Redux - Who are we?
Many of you learned about VocaliD from my TED Talk. It all began with the realization that millions of people who rely on computerized voices to communicate have access to only a handful of generic options, yet many of them have unique vocal identities that could - and should - be harnessed. We wouldn’t dream of fitting a little girl with the prosthetic limb of a grown man: why, then, the same prosthetic voice? Through scientific and technological advances we developed techniques to craft unique vocal identities by blending several hours of donor speech with as little as a vowel sound from the recipient.
What have we been up to?
At the time of the TED Talk, VocaliD was a research project; we couldn’t have predicted the overwhelming response we received from voice donors, recipients and ambassadors all over the world. Within months we had 4,000, then 10,000, and now over 25,000 supporters who have pledged to give the gift of voice. We also had more recipients sign up than we could handle.
So in May 2014 we formed VocaliD Inc., a socially motivated company aimed at designing custom-crafted voices – and we haven’t stopped working ever since. Highlights include being selected as a finalist in the MassChallenge accelerator program to help us understand how to bring this innovation to market; launching The Human Voicebank Initiative to gather the largest and most diverse repository of speech from everyday talkers; and receiving research funding from the National Science Foundation (NSF) and National Institutes on Health (NIH) to help us further our efforts to improve the quality and efficiency of creating personalized voices at minimal cost.
Our world-class, dedicated team is committed to bringing VocaliD™ voices to market by 2016. In the meantime, we are gathering donor voices through our virtual voicebank recording platform (www.VocaliD.co/voicebank), investing in continued research and development, and raising funds to support our commercialization efforts.
What’s next and how can you get involved?
You are an essential part of our solution, and without you we can’t hope to change the status quo of uniform, manufactured voices.
If you signed up to be a voice donor, please visit our virtual recording studio to begin banking your voice today www.VocaliD.co/voicebank. Many of you indicated that you had expertise to share and for that we are also grateful. To help us better understand how to engage you in our efforts, please visit us at www.VocaliD.co and click on Get Involved
where you can choose one or more of the following roles:
• Voice Donor – To bank your voice for others or for your own use
• Voice Recipient – If you or someone you know is interested in a personalized VocaliD™ voice
• Ambassador – To join our extended team and help make this a reality
• General inquiry – Stay tuned and support us from the sidelines
The next stop on our journey?
In the coming weeks you will be hearing a lot more from us as we gear up to kick off VoiceDrives across Canada and the US, increase awareness about the need for personalized voices, and launch an Indiegogo crowdfunding campaign.
We can’t do this alone! Get involved to help amplify our mission.
Sincerely,
Rupal Patel
Founder | VocaliD

Post by **Availle** » May 1st, 2015, 4:45 pm

Quoting the quote of tovarisch, adding my emphasis:

On that page they wrote:Will my recordings and information be shared with others?

We understand your personal data is sensitive information, and we make confidentiality a priority. To store and retrieve recordings and to match donor-recipient voices, we gather information such as name, age, gender, height, primary language, other language, email, etc.

What's that got to do with anything?

Isana · Post by **Isana** » May 1st, 2015, 5:11 pm

Availle wrote:Quoting the quote of tovarisch, adding my emphasis:
On that page they wrote:Will my recordings and information be shared with others?

We understand your personal data is sensitive information, and we make confidentiality a priority. To store and retrieve recordings and to match donor-recipient voices, we gather information such as name, age, gender, height, primary language, other language, email, etc.
What's that got to do with anything?

That was my first thought when I read that post.

Kangaroo692 · Post by **Kangaroo692** » May 1st, 2015, 5:15 pm

Probably to help match the generated voice with the person who needs it.

Post by **Availle** » May 1st, 2015, 5:17 pm

Why on earth is my height relevant to anybody listening to my voice?
Unless you'd want to date me?

Kangaroo692 · Post by **Kangaroo692** » May 1st, 2015, 5:20 pm

I don't really know...

Isana · Post by **Isana** » May 1st, 2015, 5:26 pm

Availle wrote:Why on earth is my height relevant to anybody listening to my voice?
Unless you'd want to date me?

I think you're on to something, Ava. I think it's really a dating service. The research is about using one's voice to find the perfect partner. There should be an app for this soon.

Kangaroo692 · Post by **Kangaroo692** » May 1st, 2015, 5:28 pm