I've had TTS Clock for a while now, and it's cool to hear it at startup. But the US female voice is starting to grate. I'm not Michael Knight, and my car is not a black sports thing! So I've installed Festival along with a British voice onto my Linux workstation. But now I need to generate the set of PCM files to go in /drive0/var/tts, and I can't find the set of inputs to Festival. I've looked on Domokun's web site, but there's nothing in the way of sources to generate the PCM files, so I'm asking here. If I have to re-do all of his work to re-generate the samples, then so be it, but I'd rather save the time for something more entertaining.

BTW, I also discovered a stray carriage-return at the end of line 19 of the script, which gives "': not a valid identifieropts: `Option" errors on the console. Removing the carriage-return makes the error message go away.

I think he used Cool Edit www.cooledit.com to convert the WAV's into PCM. Looks like he did it all by hand here. I think Cool Edit is PC only however..

If you do manage to get a british voice on all those TTS samples, would you share?

No disrespect to our American friends, but I would prefer to hear a more familiar accent first thing in the morning...

I'm guessing that sox could write them.

But the US female voice is starting to grate. I'm not Michael Knight, and my car is not a black sports thing!

KITT was voiced by William Daniels, so even if you were Michael Knight or drove a black sports car I wouldn't expect to hear a US female voice.

You could use SOX to convert WAV to PCM. But if it's already 44.1khz 16bit stereo WAVs, then you can just take out (I think) the first 64 bytes. For the most part, stripping the WAV header (whatever it is) is the same as a PCM (to my knowledge), as long as the base parameters (sampling rate, channels, etc.) are the same.

Thanks for all the advice. I don't have Windows, so CoolEdit is out of the question. I've had a play with Festival, generating WAV output and converting with sox:

all: $(subst .txt,.pcm,$(wildcard *.txt))

%.wav: %.txt
[tab] /usr/bin/text2wave $< >$@
[tab] /usr/bin/normalize $@ 2>&-


%.pcm: %.wav
[tab] /usr/bin/sox $< -t raw -r 44100 -w -c 2 $@ resample -qs

clean:
[tab] rm -f *.pcm *.wav

The speech quality is quite a long way short of the distributed TTS Clock, though - adequate for reading emails, say, but not for something I hear every day. I think I ought to investigate the alternative of persuading a friend to record the samples. I don't have studio-quality (or in fact any) recording gear, though, so it may take a while.

Domokun linked this site a while ago for generating online text-to-speech.

I had a play (so to speak) with UK-English Audrey and she sounded pretty good I even started to make replacement sound files for Domokun's TTS clock but gave up due to lack of time.

Someone should make an audio version of http://www.humanclock.com ...

Peter

Anyone remember how to convert WAV to PCM successfully under windows? Didn't have luck with sox quite yet.

Have you tried Goldwave?

Thanks Jim!

This is the closest I've gotten. PCM 32 bit Signed Stereo - but it comes out a bit sped up and higher octave than the original. 24 and 16 bit and 8 bit all sound garbled.

Downloaded it, still trying to figure out the settings for the PCM that tts/pcmplay will use. So far sounds garbled or sped up. It's been so long I forgot!