Sansa e200 Voicebox talk files

[lonerock]

Hi,
I am trying to get my Sansa e200 set up for voice.  I downloaded the latest build of RockBox, the voice file, and Voicebox.  Everything works ok except for the .talk files.  

The player won't read the .talk clips for some reason.  When I highlight the directory, the voice feature just starts to spell it and it doesn't seem to recognize the folder talk clip.

I followed the instructions to the letter.
Voicebox created the talk files successfully with no errors in the log.
Talk files are all in the right place and named correctly.  (_dirname.talk)
I turned the use talk clip options to on
I tried renaming the talk file to .mp3 and playing it through the e200, which was successful, so the encoding is ok.
I reset setting and rebooted and still doesn't read the .talk files.

I'm not sure where to look next.  Any help would be much appreciated.

[lonerock]

Sorry, I figured out my problem.  It works going through the File menu but not through the Database menu browsing by artists, album, etc.  I misunderstood that document when it stated that it would not work if browsing the database.  

That's a shame though, since that's how I would expect most people would get to the music, i.e. browsing by album or artist, instead of scrolling through hundreds of individual files.  Maybe in a future version that will be supported.

Thanks

[bascule]

Now, I'm completely unsure about this, but some of the recent commits I've seen seem to be pointing towards achieving 'dynamic' speech, using the Speex codec.

If so, I would expect that it would then be able to 'read' anything on screen, including the Database.

Remove unneeded math wrappers. Clear state structs in decoder init. Start initial work at stripping away parts of Speex in preparation for its use as a statically linked voice UI codec.

Sync Speex to SVN. Disable stereo compatibility hack since we don't needed it and it produced warnings. Remove unneeded math.h

Build libspeex for core voice on SWCODEC. Development phase - not currently utilized.

[linuxstb]

Now, I'm completely unsure about this, but some of the recent commits I've seen seem to be pointing towards achieving 'dynamic' speech, using the Speex codec.

If so, I would expect that it would then be able to 'read' anything on screen, including the Database.

"speex" is just a normal audio compression codec, similar to MP3, Vorbis etc etc.  The difference is that it's optimised for encoding spoken word content, which is why its replacing MP3 as the codec used to compress voice clips for Rockbox (on swcodec targets - Archos need to stick with mp3).

You may be confusing this with the work on "espeak" that's on flyspray - that's a "text-to-speech" plugin.