Re: [Audacity-devel] non-ascii labels

[Stuart <smc...@fr...>]

Hello Leland,

I downloaded the 1.3.1b version, and two of the problems 
still seem to be there.

> > - Editing labels using the backspace key shows that there is
> >   some byte/character confusion in the code -- one delete
> >   changes the character to something weird, a second deletes
> >   it completely.
Still a problem.
When I enter a japanese character in a label, then
move the text cursor to the end of the label and 
use the left arrow to move the cursor back over the 
characters, I have have to hit the arrow key twice
for each character.  From this (and the encoding of
exported label files, more below) I would guess that 
the label strings are encoded as cp-932 (a double-byte 
encoding and my machine's default system encoding), 
but audacity treats the string as being a single-byte 
encoding, e.g. ascii or iso-8859-x.
I guess it looks ok on the screen is because wxWidgets/
Windows don't care, and just display whatever bytes 
they're given in the default encoding.

When editing labels using the popup Edit Labels dialog,
back-arrow, character deletion, editing, etc seems to 
work fine.

> > - When an Audacity project is saved and reloaded, labels
> >   are corrupted.
Still a problem.
The labels are correctly utf-8 encoded in the .aup
file so the problem is occuring during the reading of the
.aup file.  I also noticed that the number of characters in
the audacity displayed corrupted label (determined by 
counting how many times I can use the back arrow in the 
label, since not all the corrupt characters are displayable) 
is 3 times the number of characters in the labels in the 
.aup file.  I think that all the japanese characters in my 
labels use 3 bytes when utf-8 encoded, so the problem 
may be as simple as Audacity reading the .aup file as 
a byte-encoded file rather than a utf-8 encoded one.

NB, this is only a problem with .aup files.  If I export the
labels to a file, I can import them later with no problems.
The exported label file is encoded in my machine's system
default encoding, cp-932 (a double byte encoding.)
I don't know if you see this as a problem or not.  While it
means the file can be read fine on a machine using the 
same encoding, it will likely be garbage on machines using 
a different encoding, making the exported label files non-
portable.  Maybe utf-8 would be better here too?

> > - Certain characters typed into the label editor are processed as
> >   shortcuts, causing undesired popups and preventing entry of
> >   certain characters.
This seems to be fixed.  When I enter a string that reliably
produced this effect in 1.3.0b, it works fine now.

> > - Crashes while entering/editing labels are frequent.
I did not see any crashes, even when deliberately trying
to enter or edit bad strings.

Another thing I noticed...

There is a View|Toolbars|Hide/Show TranscriptionToolbar option
but when I show it, the only thing that disappears/appears is 
a green "Play at Speed" buttom and a "Play Speed" slider.
In particular, the white box with vertical line that used to run
the Add Label at Selection action is not there.  But maybe 
work on the toolbar in not finished yet?
.
Does Audacity use unicode internally?

 HTH


land" wrote:
> Hi Stuart,
> 
> Japanese would be an awesome test of some changes we've made to the label
> handling for our upcoming 1.3.2 beta release.  If you're feeling
> adventurous, try downloading:
> 
> http://audacity.sourceforge.net/files/audacity-win-1.3.1-beta.zip
> 
> See if it fixes the problems you've described.  We'd be very happy to hear
> the results.
> 
> Thanks
> 
> Leland
> 
> On 8/17/06 6:59 PM, "Stuart" <smc...@fr...> wrote:
> 
> > Hello,
> > 
> > The recent discussions of label issues prompts me to ask
> > about some label related issues.  Please forgive and point me
> > to the right place to ask about this, if this group is not it.
> > 
> > I do (or will be doing) a lot of annotation (labeling) of recorded
> > Japanese speech, with the annotations in Japanese.  I am using
> > 1.3.0b (on Windows 2000 with a Japanese locale, cp-932 charset)
> > for this now and it sort of works but there are some annoying
> > problems:
> > - When an Audacity project is saved and reloaded, labels
> >   are corrupted.
> > - Editing labels using the backspace key shows that there is
> >   some byte/character confusion in the code -- one delete
> >   changes the character to something weird, a second deletes
> >   it completely.
> > - Certain characters typed into the label editor are processed as
> >   shortcuts, causing undesired popups and preventing entry of
> >   certain characters.
> > - Crashes while entering/editing labels are frequent.
> > 
> > I looked in Bugzilla, and didn't see anything that sounded liike
> > these issues, except for 243 [Pressing 'backspace' when adding
> > a new label crash].  Are these known issues that have been,
> > or are likely to be, fixed in the next release?
> > 
> > BTW, athough there are some problems, using Audacity to
> > do the labeling is usable.  CoolEdit 2000 fails to handle Japanese
> > characters completely.  So kudos to the Audacity developers!!