new rms voice testing #512
zstanecic
started this conversation in
Show and tell
Replies: 10 comments
-
|
Yeah, I think it's a really good voice. The only thing it needs is to be
converted to 22khz just like how slt and liubov sound. Speaking of
experiments, how long did it take for you to build the voice?
… Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Hi,
There is just a 16 k version of this, unfortunately.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 8:50 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
Yeah, I think it's a really good voice. The only thing it needs is to be
converted to 22khz just like how slt and liubov sound. Speaking of
experiments, how long did it take for you to build the voice?
Message ID: ***@***.*** <mailto:***@***.***> >
—
Reply to this email directly, view it on GitHub <#512 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACVCDE6JF2JOBOEOMVNJ7XDVA56ODANCNFSM5RF4NB4A> .
You are receiving this because you authored the thread. <https://github.com/notifications/beacon/ACVCDE224Z2RCTWBNCJYZYDVA56ODA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJSTA.gif> Message ID: ***@***.*** ***@***.***> >
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Oh I see. Why do I feal like this voice im particular is made with
clustergen instead of hts? Or is that what hts does to lower quality audios?
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Btw, to build the voice, I needed 4 hours for training.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 8:50 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
Yeah, I think it's a really good voice. The only thing it needs is to be
converted to 22khz just like how slt and liubov sound. Speaking of
experiments, how long did it take for you to build the voice?
Message ID: ***@***.*** <mailto:***@***.***> >
—
Reply to this email directly, view it on GitHub <#512 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACVCDE6JF2JOBOEOMVNJ7XDVA56ODANCNFSM5RF4NB4A> .
You are receiving this because you authored the thread. <https://github.com/notifications/beacon/ACVCDE224Z2RCTWBNCJYZYDVA56ODA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJSTA.gif> Message ID: ***@***.*** ***@***.***> >
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
so not a lot of effort for an existing text corpus along with an existing
dataset and existing language files. The trickier part is using a different
data corpus to record larger prompt selections to build possibly
higer quality voices, and the trickiest is adding a very, very very complex
language like arabic to the mix
…On Sun, 20 Mar 2022 at 22:07, Zvonimir Stanečić ***@***.***> wrote:
Btw, to build the voice, I needed 4 hours for training.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 8:50 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
Yeah, I think it's a really good voice. The only thing it needs is to be
converted to 22khz just like how slt and liubov sound. Speaking of
experiments, how long did it take for you to build the voice?
> Message ID: ***@***.*** <mailto:***@***.***> >
>
—
Reply to this email directly, view it on GitHub <
#512 (comment)>
, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/ACVCDE6JF2JOBOEOMVNJ7XDVA56ODANCNFSM5RF4NB4A>
.
You are receiving this because you authored the thread. <
https://github.com/notifications/beacon/ACVCDE224Z2RCTWBNCJYZYDVA56ODA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJSTA.gif>
Message ID: ***@***.*** ***@***.***> >
—
Reply to this email directly, view it on GitHub
<#512 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AT2FKJVXRMDJQAF2EIH3BVTVA6HS5ANCNFSM5RF4NB4A>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Yep, and Grzegorz with me, and continuous research made the polish language, i.e the first version for six months.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 11:23 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
so not a lot of effort for an existing text corpus along with an existing
dataset and existing language files. The trickier part is using a different
data corpus to record larger prompt selections to build possibly
higer quality voices, and the trickiest is adding a very, very very complex
language like arabic to the mix
On Sun, 20 Mar 2022 at 22:07, Zvonimir Stanečić ***@***.*** <mailto:***@***.***> > wrote:
Btw, to build the voice, I needed 4 hours for training.
From: king-dahmanus ***@***.*** <mailto:***@***.***> >
Sent: Sunday, March 20, 2022 8:50 PM
To: RHVoice/RHVoice ***@***.*** <mailto:***@***.***> >
Cc: Zvonimir Stanečić ***@***.*** <mailto:***@***.***> >; Author ***@***.*** <mailto:***@***.***> >
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
Yeah, I think it's a really good voice. The only thing it needs is to be
converted to 22khz just like how slt and liubov sound. Speaking of
experiments, how long did it take for you to build the voice?
> Message ID: ***@***.*** <mailto:***@***.***> <mailto:***@***.***> >
>
—
Reply to this email directly, view it on GitHub <
#512 (comment)>
, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/ACVCDE6JF2JOBOEOMVNJ7XDVA56ODANCNFSM5RF4NB4A>
.
You are receiving this because you authored the thread. <
https://github.com/notifications/beacon/ACVCDE224Z2RCTWBNCJYZYDVA56ODA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJSTA.gif>
Message ID: ***@***.*** <mailto:***@***.***> ***@***.*** <mailto:***@***.***> > >
—
Reply to this email directly, view it on GitHub
<#512 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AT2FKJVXRMDJQAF2EIH3BVTVA6HS5ANCNFSM5RF4NB4A>
.
You are receiving this because you commented.Message ID:
***@***.*** <mailto:***@***.***> >
—
Reply to this email directly, view it on GitHub <#512 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACVCDE3UUTOQH4TZTYS32E3VA6QLXANCNFSM5RF4NB4A> .
You are receiving this because you authored the thread. <https://github.com/notifications/beacon/ACVCDEYP7T7EFURROHL6NZLVA6QLXA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJV4A.gif> Message ID: ***@***.*** ***@***.***> >
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
wow that must be a lot. Maybe it'll take like 2 years or more to add arabic
On Sun, 20 Mar 2022 at 23:37, Zvonimir Stanečić ***@***.***>
wrote:
… Yep, and Grzegorz with me, and continuous research made the polish
language, i.e the first version for six months.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 11:23 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
so not a lot of effort for an existing text corpus along with an existing
dataset and existing language files. The trickier part is using a different
data corpus to record larger prompt selections to build possibly
higer quality voices, and the trickiest is adding a very, very very complex
language like arabic to the mix
On Sun, 20 Mar 2022 at 22:07, Zvonimir Stanečić ***@***.*** <mailto:***@***.***>
>
wrote:
> Btw, to build the voice, I needed 4 hours for training.
>
>
>
> From: king-dahmanus ***@***.*** <mailto:***@***.***> >
> Sent: Sunday, March 20, 2022 8:50 PM
> To: RHVoice/RHVoice ***@***.*** <mailto:***@***.***> >
> Cc: Zvonimir Stanečić ***@***.*** <mailto:***@***.***> >; Author
***@***.*** <mailto:***@***.***> >
> Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
>
>
>
> Yeah, I think it's a really good voice. The only thing it needs is to be
> converted to 22khz just like how slt and liubov sound. Speaking of
> experiments, how long did it take for you to build the voice?
>
> > Message ID: ***@***.*** <mailto:***@***.***> <mailto:***@***.***> >
> >
>
>
>
> —
> Reply to this email directly, view it on GitHub <
>
#512 (comment)
>
> , or unsubscribe <
>
https://github.com/notifications/unsubscribe-auth/ACVCDE6JF2JOBOEOMVNJ7XDVA56ODANCNFSM5RF4NB4A
>
> .
> You are receiving this because you authored the thread. <
>
https://github.com/notifications/beacon/ACVCDE224Z2RCTWBNCJYZYDVA56ODA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJSTA.gif
>
> Message ID: ***@***.*** <mailto:***@***.***> ***@***.*** <mailto:***@***.***>
> >
>
> —
> Reply to this email directly, view it on GitHub
> <
#512 (comment)
>,
> or unsubscribe
> <
https://github.com/notifications/unsubscribe-auth/AT2FKJVXRMDJQAF2EIH3BVTVA6HS5ANCNFSM5RF4NB4A
>
> .
> You are receiving this because you commented.Message ID:
> ***@***.*** <mailto:***@***.***> >
>
—
Reply to this email directly, view it on GitHub <
#512 (comment)>
, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/ACVCDE3UUTOQH4TZTYS32E3VA6QLXANCNFSM5RF4NB4A>
.
You are receiving this because you authored the thread. <
https://github.com/notifications/beacon/ACVCDEYP7T7EFURROHL6NZLVA6QLXA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJV4A.gif>
Message ID: ***@***.*** ***@***.***> >
—
Reply to this email directly, view it on GitHub
<#512 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AT2FKJWVV5XD5MG2WKIYSEDVA6SD3ANCNFSM5RF4NB4A>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
also, I thought that this rhvoice had the same feature as festvox where you
could make the voice sound as high quality as 22k, even if you have an only
sixteen k dataset. Guess I shouldn't do these comparisons a lot lolrofl
…On Sun, 20 Mar 2022 at 21:32, blind zigzigon ***@***.***> wrote:
Oh I see. Why do I feal like this voice im particular is made with
clustergen instead of hts? Or is that what hts does to lower quality audios?
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
No, this HTS does to the lower quality audio.
From: king-dahmanus ***@***.***>
Sent: Sunday, March 20, 2022 9:33 PM
To: RHVoice/RHVoice ***@***.***>
Cc: Zvonimir Stanečić ***@***.***>; Author ***@***.***>
Subject: Re: [RHVoice/RHVoice] new rms voice testing (Discussion #512)
Oh I see. Why do I feal like this voice im particular is made with
clustergen instead of hts? Or is that what hts does to lower quality audios?
—
Reply to this email directly, view it on GitHub <#512 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACVCDE5W65UEUXRVEJ4VWQ3VA6DPZANCNFSM5RF4NB4A> .
You are receiving this because you authored the thread. <https://github.com/notifications/beacon/ACVCDEZEKZIMVPVHFMGYJNTVA6DPZA5CNFSM5RF4NB4KYY3PNVWWK3TUL52HS4DFWFCGS43DOVZXG2LPNZBW63LNMVXHJKTDN5WW2ZLOORPWSZGOAASJT2I.gif> Message ID: ***@***.*** ***@***.***> >
|
Beta Was this translation helpful? Give feedback.
0 replies
-
|
wait, what do you mean? I didn't completely understand your statement |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi!
This is the experiment version of the RMS voice. Who can tel me, is it a really good thing? Is it worth it? Rename .zip to nvda-addon to test it with NVDA>
RHVoice-voice-English-Rms.zip
Beta Was this translation helpful? Give feedback.
All reactions