6 Feb
2017
6 Feb
'17
10:26 p.m.
Hi everyone,
this is a script to make a dramatised reading from a plain text file, using
speech synthesizers on both Linux and MAC OS. Hear the software play it for
you. :)
https://freeshell.de/~silvain/software/story_maker-1.0rc1.tar.bz2
It works from the Linux terminal and only needs sndfile (sndfile-concat and
sndfile-convert) and optionally oggenc or lame to create a fully tagged
audiobook.
story_maker can use sapi4linux (sapilektor), espeak and festival on Linux and
the standard MAC OS voices (through say), which includes the InfoVox IVox
voices.
Features:
automatically find direct speech
interactively assign character names to direct speech passages
choose a cast of voices for your characters
save a full or partial cast and use it for other stories
automatically render on Linux and (optionally) MAC (using ssh)
optionally build a fully tagged OGG or mp3 file
I can't publicly supply my demo, since it uses proprietary voices, not meant
for published audiofiles. :( But it sounds OK. It's certainly better than
reading it at the computer scren or braille display and sounds much better
than plain old espeak. :)
I'd love to hear some feedback if you go for that crazy thing. <3
Best wishes,
Jeanette
--------
When you need someone, you just turn around and I will be there <3
7 Feb
7 Feb
12:23 p.m.
New subject: Release: story_maker - make dramatised readings with your speech synths
On 06 Feb 2017, at 22:26, Jeanette C.
<julien(a)mail.uni-paderborn.de> wrote:
I can't publicly supply my demo, since it uses proprietary voices, not meant for
published audiofiles.
Are you sure that it isn't legal to use proprietary voices? If we pay for e.g. a
proprietary orchestra sample library, it wouldn't make sense, if it would be illegal
to use the orchestra sounds for publishing.
--
"Martin Schulz rockt die SPD", sowie Stromberg als Nachfolger von Lemmy frischen
Wind in Motörhead bringen könnte. Wenn man Schulz hört und/ oder sieht, ist die
Assoziation das er "rockt"? Wer das Wort "rocken" in diesem
Zusammenhang benutzt, hat offensichtlich keine Ahnung vom Leben.
12:30 p.m.
Feb 7 2017, Ralf Mardorf has written:
...
Are you sure that it isn't legal to use
proprietary voices? If we pay for e.g. a proprietary orchestra sample library, it
wouldn't make sense, if it would be illegal to use the orchestra sounds for
publishing.
Slightly different case, those voices are really cheap ($15 and for
some
other $45). They sell other licenses to the industry for voice branding.
So, these voices are meant for personal use only.
Anyway, I just pushed an update, with a few improvements and two
examples using espeak. :)
https://freeshell.de/~silvain/software/story_maker-1.0rc2.tar.bz2
Byebye,
Jeanette
...
--------
When you need someone, you just turn around and I will be there <3
10:24 p.m.
New subject: Release: story_maker - make dramatised readings with your speech synths
Jaeanette:
...
Anyway, I just pushed an update, with a few
improvements and two
examples using espeak. :)
https://freeshell.de/~silvain/software/story_maker-1.0rc2.tar.bz2
README, is this a typo:
* If possible aotumatically guess the title of the text
If I understand your texts and code, your code is about finding
passages, identify persons that should speak thoose passages and choose
voices so that a each person has its own voice-sound. Do you try to
identify gender, old persons, children or persons with dialect, etc.
in the texts also ?
Regards,
/Karl Hammar
-----------------------------------------------------------------------
Aspö Data
Lilla Aspö 148
S-742 94 Östhammar
Sweden
+46 173 140 57
10:35 p.m.
Feb 7 2017, karl(a)aspodata.se has written:
...
README, is this a typo:
* If possible aotumatically guess the title of the text
Thanks, of course it is.
:)
If I understand your texts and code, your code is about finding
passages, identify persons that should speak thoose passages and choose
voices so that a each person has its own voice-sound.
Yes.
Do you try to
identify gender, old persons, children or persons with dialect, etc.
in the texts also ?
No, not as such. I don't have too many voices anyway. At
least not ones
that sound really good. I have one genuine child's voice and a few adult
voices, who could be of different ages. I just try to find the right one
for each character. Naturally I often have to use voices for multiple
characters. So characters with few lines might share voices. I try to
fit the main characters with different voices and not re-use voices for
main parts. But it's partly a question of necessity and resources. I
suppose that in English - accepting some or all dialects - there's a
better choice. In German I could theoretically use about 12 or 13
acceptable and good voices.
I hope you can get some enjoyment out of this. :)
Due to other comments, I'm planning of rewriting it in c++. Requested
features included a sort of dictionary or language learning scenario and
reading HTML texts, using different voices not for characters, but to
mark headings and other text attributes or container types. The latter
one doesn't sound like something I'd do very soon, but it sounds useful.
Best wishes,
Jeanette
...
--------
When you need someone, you just turn around and I will be there <3
11:08 p.m.
New subject: Release: story_maker - make dramatised readings with your speech synths
Jeanette:
...
Feb 7 2017, karl(a)aspodata.se has written:
...
Do you try to
identify gender, old persons, children or persons with dialect, etc.
in the texts also ?
No, not as such. I don't have too many voices anyway. At
least not ones better choice. In German I could theoretically use
about 12 or 13
acceptable and good voices.
Ok, I see. I would be interesting if there were any voices that you
could give a parameter like angry, age and such, but I guess there
arn't any.
I hope you can get some enjoyment out of this. :)
Yea, a little...
Due to other comments, I'm planning of rewriting
it in c++. Requested
features included a sort of dictionary or language learning scenario and
reading HTML texts, using different voices not for characters, but to
mark headings and other text attributes or container types. The latter
one doesn't sound like something I'd do very soon, but it sounds useful.
Good luck then!
Regards,
/Karl Hammar
-----------------------------------------------------------------------
Aspö Data
Lilla Aspö 148
S-742 94 Östhammar
Sweden
+46 173 140 57
11:13 p.m.
Feb 7 2017, karl(a)aspodata.se has written:
...
Ok, I see. I would be interesting if there were any
voices that you
could give a parameter like angry, age and such, but I guess there
arn't any.
If they are of good quality, not as a parameter. But there are some
story teller or emotional voices with InfoVox. They have something like
happy, sad, whispering, shouting )from afar), bad guy or old man. But
there aren't many voices which can do all of that. Though with voices
like these, a different kind of tagging might be done for a story, not
as a sort of dramatised reading, but a good natural reading with one
actor.
...
Best wishes,
Jeanette
--------
When you need someone, you just turn around and I will be there <3
2850
days inactive
2851
days old
linux-audio-user@lists.linuxaudio.org
6 comments
3 participants
participants (3)
-
Jeanette C. -
karl＠aspodata.se -
Ralf Mardorf