23 Nov
2005
23 Nov
'05
3:48 p.m.
Hello, all,
Does anyone know if there's a way to convert text files to MP3s
in Linux? I'd like to be able to take some of my text files and make
audio out of them so I can listen to them while walking.
Thanks,
Terrence
5:02 p.m.
Festival is good for this, as well as embrola (which somebody else
already suggested).
Dave
On 11/23/05, Terrence <terrencevak-lau(a)terrencevak.net> wrote:
Hello, all,
Does anyone know if there's a way to convert text files to MP3s
in Linux? I'd like to be able to take some of my text files and make
audio out of them so I can listen to them while walking.
Thanks,
Terrence
6:10 p.m.
Terrence <terrencevak-lau(a)terrencevak.net>et>:
Hello, all,
Does anyone know if there's a way to convert text files to MP3s
in Linux? I'd like to be able to take some of my text files and make
audio out of them so I can listen to them while walking.
"mbrola" (free-beer license, closed source from academic sector) takes a
list of phonemes as input to create audio output. So you need a
language specific text-to-phonemes tool. The mbrola site has links for
those (festival, txt2pho etc.). To turn textfiles into ogg/mp3 you
would setup a chain of shell pipes like
cat textfile | txt2pho | mbrola voicefile - - | encoder
http://tcts.fpms.ac.be/synthesis/mbrola.html
Are you a noisecore person? You'd then just get rid of parts 2 and 3 of
that very chain of course.
Wolfgang
6:48 p.m.
Wolfgang Woehl writes:
cat textfile | txt2pho | mbrola voicefile - - |
encoder
...
Are you a noisecore person? ...
I'm not familiar with noisecore. Is that another encoder?
I tried doing:
cat an.txt | festival --tts | mbrola test.mp3 - - | lame -
but it just did a little encoding that sent the results to the screen.
Am I missing something, or misinterpreting something?
Thanks,
Terrence
7:42 p.m.
Terrence <terrencevak-lau(a)terrencevak.net>et>:
Wolfgang Woehl writes:
^^^^^^^^
Syntax is "mbrola <path to voicefile> <phoneme file>
<output>"
So you would say something like
... | mbrola en1 - - | ...
cat textfile | txt2pho | mbrola voicefile - - |
encoder
I tried doing:
cat an.txt | festival --tts | mbrola test.mp3 - - | lame - but it just did a little encoding that sent the
results to the
screen. Am I missing something, or misinterpreting something?
I don't know. Me not using lame cause mp3 so lame. But you will probably
have to tell it that the input is raw pcm, maybe the samplerate that
comes out of mbrola and a name for the file lame is supposed to lame,
err, create. Manpage has details, I'm so sure :)
Wolfgang
24 Nov
24 Nov
1:05 a.m.
Which does the "voice file" and "phoneme file" refer to? Are
those the text and MP3 files?
Thanks,
Terrence
6:32 a.m.
Terrence <terrencevak-lau(a)terrencevak.net>et>:
Which does the "voice file" and
"phoneme file" refer to? Are
those the text and MP3 files?
Terrence, there really is no way around reading up on the mbrola stuff.
Read "mbrola -h" and the webpage. You need a "voice" so that mbrola
can
say something. See the mbrola pages for links to voices. All this is
not related to mp3 at all.
The basic dataflow is
1. Transform a textfile into a list of phonemes that
2. mbrola understands and turns into audio data which
3. You can then pick up and do with it whatever
Wolfgang
10:15 a.m.
Wolfgang Woehl writes:
Terrence <terrencevak-lau(a)terrencevak.net>et>:
OK, from reading what you've said, and from reading the MBROLA web page,
it seems I need an actual phoneme database file. I'm on the synthesis
page, but only see a bunch of .wav file links and a nav bar. Am I
looking in the wrong place, or looking at it wrong? Also, what is
noisecore?
Thanks,
Terrence
Which does the "voice file" and
"phoneme file" refer to? Are
those the text and MP3 files?
Terrence, there really is no way around reading up on the mbrola stuff.
Read "mbrola -h" and the webpage. You need a "voice" so that mbrola
can
say something. See the mbrola pages for links to voices. All this is
not related to mp3 at all.
The basic dataflow is
1. Transform a textfile into a list of phonemes that
2. mbrola understands and turns into audio data which
3. You can then pick up and do with it whatever
Wolfgang
11:31 a.m.
Terrence wrote:
Also, what is noisecore?
It's a style of music, generally involving the creation of lots of
distorted sounds & noise.
Basically, the original use of the word in this thread was because the
guy was trying to account for all possibilities...
He was wondering if you wanted to convert the text directly to audio -
without using conventional speech synthesis techniques. It's just
another interesting way of creating sound.
I guess it may have been partly in jest.
Regards
Michael Nelson
1:47 p.m.
Gotcha! Yeah, I've done that just to see what would happen,
taking a text file and "cat"ing it out to /dev/dsp. I've actually sort
of produced some funky broken chords doing that.
Terrence
5:01 p.m.
OK, I got a phoneme database, but how can I create the .pho from
my text file?
Thanks again,
Terrence
6:50 p.m.
Terrence <terrencevak-lau(a)terrencevak.net>et>:
OK, I got a phoneme database, but how can I create
the .pho from
my text file?
I only know txt2pho which is for german text. You seem to have festival
and will have to figure it out. From skimming over the festival stuff I
take it that festival might not even need mbrola. Festival's manpage
will help, ok?
Wolfgang
25 Nov
25 Nov
9:58 a.m.
On Thursday 24 November 2005 22:41, Wolfgang Woehl was like:
I only know txt2pho which is for german text. You seem
to have festival
and will have to figure it out. From skimming over the festival stuff I
take it that festival might not even need mbrola. Festival's manpage
will help, ok?
Festival doesn't require mbrola AFAIK.
Neither does flite
--
cheers,
tim hall
http://glastonburymusic.org.uk/tim
26 Nov
26 Nov
3:58 a.m.
tim hall writes:
Festival doesn't require mbrola AFAIK.
Neither does flite
Nope, they don't. `-) Flite is what I'm using because it definitely
supports wave-out. Festival, from what I've heard and experienced, only
outputs to the sound card and nothing else.
Thanks, guys!
Terrence
8:39 a.m.
Terrence <terrencevak-lau(a)terrencevak.net> writes:
Festival, from what I've heard and experienced,
only outputs to the
sound card and nothing else.
Oh no it doesn't! Put something like this in your .festivalrc:
(Parameter.set 'Audio_Command "sox -r 16000 -w -s -t raw $FILE -t wav
festival-out.wav")
(Parameter.set 'Audio_Method 'Audio_Command)
Then "echo hello world | festival --tts" will result in it creating
festival-out.wav with that phrase in it.
--
Adam Sampson <ats(a)offog.org> <http://offog.org/>
6735
days inactive
6738
days old
linux-audio-user@lists.linuxaudio.org
15 comments
7 participants
participants (7)
-
Adam Sampson -
Christoph Eckert -
David Jaquay -
Michael T D Nelson -
Terrence -
tim hall -
Wolfgang Woehl