Description: A version of Nighto's amarokTalker that instead uses Espeak, as it's much faster than festival.
Basically it will announce the currently playing track, using espeak to generate a .wav file (put into the /tmp directory), and then aplay to play it.Last changelog:
v0.2 29/08/2007 Now says the name of the album, if it's different from the last track. Also added checks for the aplay and espeak binaries.
Looking at the code, i thought about something.
I don't know if this is the default, but at least in the distro i'm using (Ubuntu) the /tmp folder is flushed in every boot. So, IMO, you shouldn't worry about erasing the wav files.
Am I wrong? Why?
Thinking about the caching option, we should think of naming the wav files without random numbers, but with the title/+artist of the music. Going further, even (after playing) encoding the wav file into a (low bitrated) ogg, and before playing, check if a ogg file named accordingly exists.
Just some thoughts. This sadly is over my scripting capacities =/
Waiting for next version ;)
Espeak is realy better than festival. I found buggy script on kde-apps.org (not AmarokFestival), that used festival, modified for espeak and fixed bugs.
I do not downloaded your script, but have a question: Do you kill the espeak before saying track?
On fast track change maybe multiple instances of espeak and you didn't hear anything. I do `killall espeak` before launching
Sorry for my English
I use espeak to generate a .wav file in the /tmp directory, then use aplay to play it, as by default espeak tries to open the /dev/dsp device (which of course is locked when amarok is playing).
I don't have espeak running all the time, just start it to generate the wav file, after which it exits.
My script uses libaoss.so (alsa-oss package)
So, to start espeak I need to run:
killall espeak
LD_PRELOAD=libaoss.so espeak "text to say"
It works, and I think it will be better than generate a wav file
I had a look at that method on my system this evening, but unfortunately all that happens is that espeak pushes the CPU usage to 100% without actually producing any speech (probably a misconfiguration on my Ubuntu 7.04 system). In the end I give up waiting for it to complete, and terminate the process.
Thanks for the suggestion though!
Wow! I didn't know espeak... it sounds (IMO) way better than festival. It also supports my native language (pt-br).
I noticed espeak supports lots of languages, btw, and wondered about a configuration window for choosing the language, but I have no idea of how to do it.
Maybe that window should have the option of choosing or not the cache activation, because each wav file is about 200kB. For those who have a big music collection, that may be bad.
[]
Nighto
I've tried a quite a few tracks now, and I think you're right, that Espeak is clearer than festival.
I'm toying with the idea of rewriting the script in python, along with the inclusion of configuration options, so watch this space!
Ratings & Comments
8 Comments
Looking at the code, i thought about something. I don't know if this is the default, but at least in the distro i'm using (Ubuntu) the /tmp folder is flushed in every boot. So, IMO, you shouldn't worry about erasing the wav files. Am I wrong? Why? Thinking about the caching option, we should think of naming the wav files without random numbers, but with the title/+artist of the music. Going further, even (after playing) encoding the wav file into a (low bitrated) ogg, and before playing, check if a ogg file named accordingly exists. Just some thoughts. This sadly is over my scripting capacities =/ Waiting for next version ;)
Cool! Rock on!
Espeak is realy better than festival. I found buggy script on kde-apps.org (not AmarokFestival), that used festival, modified for espeak and fixed bugs. I do not downloaded your script, but have a question: Do you kill the espeak before saying track? On fast track change maybe multiple instances of espeak and you didn't hear anything. I do `killall espeak` before launching Sorry for my English
I use espeak to generate a .wav file in the /tmp directory, then use aplay to play it, as by default espeak tries to open the /dev/dsp device (which of course is locked when amarok is playing). I don't have espeak running all the time, just start it to generate the wav file, after which it exits.
My script uses libaoss.so (alsa-oss package) So, to start espeak I need to run: killall espeak LD_PRELOAD=libaoss.so espeak "text to say" It works, and I think it will be better than generate a wav file
I had a look at that method on my system this evening, but unfortunately all that happens is that espeak pushes the CPU usage to 100% without actually producing any speech (probably a misconfiguration on my Ubuntu 7.04 system). In the end I give up waiting for it to complete, and terminate the process. Thanks for the suggestion though!
Wow! I didn't know espeak... it sounds (IMO) way better than festival. It also supports my native language (pt-br). I noticed espeak supports lots of languages, btw, and wondered about a configuration window for choosing the language, but I have no idea of how to do it. Maybe that window should have the option of choosing or not the cache activation, because each wav file is about 200kB. For those who have a big music collection, that may be bad. [] Nighto
I've tried a quite a few tracks now, and I think you're right, that Espeak is clearer than festival. I'm toying with the idea of rewriting the script in python, along with the inclusion of configuration options, so watch this space!