ReVoxer - the LibriVox Recorder Development Proj

Post by **hugh** » January 16th, 2006, 9:13 am

ReVoxer development project
ReVoxer is a software tool designed for recording human-read audio from text. The tool includes the functionality of audio software such as Audacity, with features tailored for text-to-audio recordings, including:
*text-scrolling which feeds text to the screen as the reader reads.
*text-audio synchronization allowing for text-string searches of audio.

Umut Topkara is looking for some help on the project (GPL of course). Post here if you are interested.

project homepage:
http://www.cs.purdue.edu/homes/utopkara/ReVoxer.html

historical threads:
http://librivox.org/forum/viewtopic.php?t=945
http://librivox.org/forum/viewtopic.php?t=660

tshirt · Post by **tshirt** » January 16th, 2006, 3:25 pm

All kinds of help are welcome from LibriVox users.

You can contribute with:
* sharing your experience with your current recording/listening program
* adding to/refining design of ReVoxer
* writing code
* creating a webpage design
* designing a logo

Post by **hugh** » January 16th, 2006, 3:51 pm

hey tshirt - we'll likely have to go outside the LV family to get actual coders, but i think you have plenty of guinnea pigs here to do the testing of alpha/beta versions...

as for the other things we might get some bites!

the main problem with audacity is: crashes!

LibraryLady · Post by **LibraryLady** » January 16th, 2006, 4:04 pm

Okay, this is all pretty much over my head but I'm certainly willing to be a guinea pig and try this fancy new idea out when it is up and running. I use Audacity and it works pretty well for me. One thing I'd like, which I know someone else already mentioned in the original thread, is a bookmark feature. That way when I'm recording and I goof, I put a bookmark and then go back and re-read the section. That would make editing easier to pick out those goofs that need to be cut.

tshirt · Post by **tshirt** » January 16th, 2006, 4:21 pm

hugh wrote:the main problem with audacity is: crashes!

This is a good reason to target a standalone program
rather than a plugin.

One nice other thing about audacity is that it's available
for many platforms. I think we should take this
one step further, and make ReVoxer run from a web
browser. This will not only make it available to many
platforms, but also make it easy to create text content
coupled with audio. The ultimate ReVoxer will be one
like the BBCode editor that I am using now to enter this
post.

tshirt · Post by **tshirt** » January 16th, 2006, 4:26 pm

LibraryLady wrote:One thing I'd like, which I know someone else already mentioned in the original thread, is a bookmark feature. That way when I'm recording and I goof, I put a bookmark and then go back and re-read the section. That would make editing easier to pick out those goofs that need to be cut.

Hi Annie,
we were wondering whether your audio bookmarks correspond to
sentence boundaries, or paragraph boundaries, or somewhere else
in the text. could you give us some feedback about where you
usually put your bookmarks?

Post by **hugh** » January 16th, 2006, 4:38 pm

oh so recording from a browser. that's a very cool idea - but it's going to be hard to get consistent audio quality: with network slowness etc. my internet cuts out occasionally. and one thing is certain when you are recording a text, your recording software crashing is so unbelieveably terrible.

have you seen;

www.podomatic.com

?

tshirt · Post by **tshirt** » January 16th, 2006, 4:47 pm

hugh wrote:oh so recording from a browser. that's a very cool idea - but it's going to be hard to get consistent audio quality: with network slowness etc. my internet cuts out occasionally.

Think of it as if the BBCode editor allowed you to save your work
on your own disk besides submitting over the internet. There are
many technical issues involved, but the general idea is having
the flexibility to move freely within the spectrum of options that lie
between saving everything on the disk and saving everything over
the network.

LibraryLady · Post by **LibraryLady** » January 16th, 2006, 4:49 pm

tshirt wrote:Hi Annie,
we were wondering whether your audio bookmarks correspond to
sentence boundaries, or paragraph boundaries, or somewhere else
in the text. could you give us some feedback about where you
usually put your bookmarks?

Right now, my "bookmarks" consist of pausing for about ten seconds so that when I look at the sound file in Audacity, I see a blank spot and know that is where I need to edit. Usually these correspond with sentence boundaries. I goof something up, pause ten seconds, then start the sentence over. Does that help?

tshirt · Post by **tshirt** » January 16th, 2006, 4:54 pm

hugh wrote:have you seen;
www.podomatic.com
?

Awesome! Does it save everything over the web?
answer= Yes

Interface is perfectly simple!

This is very close to what we should target. But we
need to have the ability to save everything to local
disk. People will probably would like to work on their
recordings before submitting them; and of course
there is the connection drop issue.

But definitely this is close to it.

tshirt · Post by **tshirt** » January 16th, 2006, 4:58 pm

LibraryLady wrote:Does that help?

Yes, very helpful.

As a hint: we were discussing with hugh about providing them book texts
such that every line will have one sentence. Once this is done we will use
sentences as atomic segments/chunks of recorded audio.

tshirt · Post by **tshirt** » January 16th, 2006, 5:01 pm

hugh wrote: www.podomatic.com
?

Another issue is macromedia flash.
I am a bit concerned that flash is not that
portable. I think we could achieve the
same functionality with java.

Post by **hugh** » January 16th, 2006, 5:38 pm

also: sound quality is crap.

tshirt · Post by **tshirt** » January 17th, 2006, 1:07 pm

hugh wrote:also: sound quality is crap.

Most probably because it saves everything over the
internet, and has to use a brutal lossy compression.

This raises other questions to ask LibriVox volunteers
and listeners:

* What compression algorithm and settings (bitrate?) do
LibriVox volunteers use?

* Which ones are preferred by listeners? Does anybody
re-encode the audio?

Speex, mp3, ogg?

I assume that the largest possible bitrate is the most
appropriate for recording; since we could always
downgrade the quality of the audio afterwards.

However this might change if we extend the usage of ReVoxer
to other applications (forum posts in audio, podcasts,
manuals, cookbooks, etc); and the listener's choice in those
applications will determine the most appropriate audio
encoding.

tshirt · Post by **tshirt** » January 18th, 2006, 1:41 pm

FYI: we have a sourceforge project now

http://sourceforge.net/projects/revoxer/

and a new project page

http://revoxer.sourceforge.net