[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Mozilla TTS (was Re: New Nuance Conversational Voice)



There is interesting text to speech work underway at Mozilla - all free software/open-source.

Demonstration: https://soundcloud.com/user-565970875/commonvoice-loc-sens-attn

Git repository: https://github.com/mozilla/TTS

If it becomes reliable and works in real time, it could be adopted as the default for Linux systems.

On 1/10/20, 11:04, "Devin Prater" <r.d.t.prater@gmail.com> wrote:

    I cannot confirm this, but there is the point that screen readers only send one sentence at a time to the speech synthesizer. Now, what happens with Emacspeak or other more regular TTS tools I don’t know, but the fact that speech synthesizers sound pretty much the same, besides Macintalk, when speaking text with screen readers and when using TTS readers, proves pretty soundly to me that almost all speech synthesizers do not read further than a sentence.
    
    > On Jan 10, 2020, at 9:53 AM, D.J.J. Ring, Jr. <n1ea@arrl.net> wrote:
    > 
    > Devin,
    > 
    > Not to go too far off topic with this, but how does it speak only one
    > sentence at a time.
    > 
    > I have nothing to use it on right now, so I don't know.
    > 
    > Can anyone confirm that this one and only "conversational" voice just
    > speak one sentence at a time?
    > 
    > Agree, that's not conversational, conversations occur in paragraphs,
    > even if it's a one word paragraph, like "Good Grief, Charlie Brown."
    > 
    > Thanks,
    > 
    > David
    > 
    > On Fri, Jan 10, 2020 at 10:16 AM Devin Prater <r.d.t.prater@gmail.com> wrote:
    >> 
    >> It’s okay, probably the best Nuance has, but they’ll need to do more than just go one sentence at a time in order to be “conversational.” I think what they mean by that is that the voice sounds smooth and pretty, for their actual customers now, the automated phone software market. Don’t be surprised if this voice doesn’t sound as great when dealing with long texts like books which we use, and even screen reading, with our jargon like “combo box” or “radio button.”
    >> 
    >> Just trying to be the voice of some reason here.
    >> 
    >> On Jan 10, 2020, at 8:45 AM, D.J.J. Ring, Jr. <n1ea@arrl.net> wrote:
    >> 
    >> Everyone,
    >> 
    >> There is only one "conversational" voice at least in any English that I found.
    >> 
    >> I was blown away when I listened to it.
    >> 
    >> It's fabulous.
    >> 
    >> It's called Zoe conversational.
    >> 
    >> https://www.nuance.com/omni-channel-customer-engagement/voice-and-ivr/text-to-speech.html
    >> 
    >> Can Voxin get permission for this? It's not mentioned, but it might be new.
    >> 
    >> After you hear it, you'll love it.
    >> 
    >> I want it so much! Gilles, please, pretty please with chocolate croissant and café au lait, please!
    >> 
    >> Best
    >> David
    >> 
    >> 
    
    



Reply to: