Would you like to mentor a GSoC student to work on chapter reading support in TagLib? (originally for Amarok)

Sat May 4 16:00:39 UTC 2013

Hi TagLib developers,
Google Summer of Code student proposal deadline was yesterday and we're quite 
happy that we've got a fair number of good proposals in Amarok. On the other 
hand we unfortunately don't have enough mentoring manpower to mentor all the 
good proposals (doesn't necessary mean this one).

One of the proposals is "Implementing CUE sheet and chapter support in Amarok" 
by Abhinandan Ramprasath, which may (despite the title) touch more TagLib than 
Amarok.

Perhaps one of you (Lukáši?) would be able to help us (and TagLib) by 
mentoring this proposal? Some reasons supporting it:
 * most solid bits of the proposal are IMO the ones touching actual metadata 
parsing, parts about Amarok integration are more brief. Student's past 
involvement touched also the metadata reading bits.
 * chapter reading from MP4 (QuickTime and Nero subformats), Ogg files and 
embedded CUE sheet reading from FLAC (and perhaps wavpack) files would be of 
good benefit to whole TagLib community, not just Amarok.
 * Amarok integration points in the proposal could be replaced by implementing 
cue or chapter reading from more file formats (already mentioned above) as 
already outlined in the proposal, should be student agree on this.

Formally, applications for Amarok are under KDE and given TagLib's strong ties 
with KDE, this shouldn't be a problem. OTOH mentoring is quite a commitment 
with reports that it takes ~1 hour a working day during the coding period 
(June 17 - September 29). [3][4]

What do you think? We'd be glad to receive any indication from you by this 
Monday (as it is the date we have to coordinate GSoC proposals within KDE)

Regards,
	Matěj Laitl of Amarok

-----------
Original idea at [1], discussion with student about the proposal at [2], copy 
of the proposal follows.

[1] 
http://community.kde.org/GSoC/2013/Ideas#Project:_Unified_CUE_File_and_Audiobook_Chapter_Support_in_Amarok
[2] https://mail.kde.org/pipermail/amarok-devel/2013-April/012031.html and 
https://mail.kde.org/pipermail/amarok-devel/2013-May/012078.html
[3] https://code.google.com/p/google-summer-of-code/wiki/AdviceforMentors
[4] http://en.flossmanuals.net/gsocmentoring/

On 3. 5. 2013 Abhinandan Ramprasath <abhiin1947 at gmail.com> wrote:
> This is the proposal I'm submitting for GSoC. Thanks to Matej for the
> reviews.
> 
> *Short Description:*
> 
> Amarok currently has very limited CUE sheet support
> and no support for chapters. My project aims at detecting chapter markers
> of various formats by implementing them in TagLib and using it to Amarok's
> advantage. Also, Integrating CUE sheets with Amarok so that they work
> seamlessly.
> 
> Basic Details:
> 
> Name: Abhinandan Ramaprasath
> Email Address: abhiin1947 at gmail.com
> Freenode IRC Nick: abhiin1947
> IM Service and Username: Gtalk - abhiin1947
> Location (City, Country and/or Time Zone): Chennai, India, UTC+05:30
> Github: https://github.com/abhiin1947/
> Proposal Title: Implementing CUE sheet and chapter support in Amarok
> 
> Motivation for project:
> 
> Audiobooks are usually quite lengthy and go on for hours. It is not always
> possible for people to finish listening to audiobooks in one go. Chapter
> support is an important feature in modern audio players. Trying to find the
> part of the audiobook that the user wants is an arduous task. Since Amarok
> already provides a bookmark to the location where the user left off,
> addition of chapter markers would make navigating an audiobook as easy as
> flipping through a book.
> 
> Chapter markers otherwise have been a demanded feature ( with over 40+
> wishes ). It would be very useful in enhanced podcasts
> <http://en.wikipedia.org/wiki/Podcast#Enhanced_podcasts> to be able
> to jump to various chapters/parts with just a click of a button.
> 
> CUE sheeted music, on the other hand, is very popular, and not providing
> proper support ruins the music experience. It has also been a requested
> feature for quite a long time (1) ( 700+ votes ). Many bugs have been
> reported (2) about Amarok’s lack of support for CUE sheeted music.
> Implementation of CUE sheet support has been delayed over releases (from
> 2.6 to 2.8) and I would like to complete it this Google Summer of Code.
> 
> Goals:
> 
> The goals of my project would be to
> 1) Implement QuickTime chapter support in TagLib.
> 2) MP3 chapter support in Amarok.
> 3) Implement OGG chapter support in TagLib.
> 4) Display these chapters in Amarok.
> 5) Integrate CUE sheet support with collection manager and Amarok for a
> seamless music experience.
> 6) Implement support for NeroChapter type in TagLib
> 7) Solve bugs related to my project along the way (3).
> 
> Implementation Details:
> 
> The project involves making changes in both TagLib and Amarok repositories.
> I would like to divide the project into 5 parts.
> 
> *1. QtChapter support in MP4/M4A/M4B files*
> 
> The first part is implementing the QtChapter (Qt = QuickTime in this
> proposal) support in MP4/M4A/M4B files. As per the Apple
> docs<https://developer.apple.com/library/mac/#documentation/QuickTime/qtff/Q
> TFFChap2/qtff2.html>, the Qt (QuickTime) chapter support can be achieved by
> reading the “trak” atom which contains the “chap” data. This trak atom
> contains the text track which contains the chapter information. The
> indicated text track should contain a “stbl” atom ( basically a table )
> which would contain offsets to the location in the file that contain
> chapter title in the file. Reading data at these offsets give information
> about chapter titles. The time at which each chapter marker occurs can be
> easily calculated using the “bitrate” of the audiofile. These would be the
> changes that I would make in TagLib to support QtChapters (QuickTime).
> 
> This should also solve part of the bug
> <https://bugs.kde.org/show_bug.cgi?id=144849>. This merge will provide
> support for the Apple Enhanced Podcast ( M4A ) as well.
> 
> The changes in Amarok include, displaying the chapter markers to the user
> in the form of bookmarks or Multi-Tracks.I plan on reading the chapter
> information in Amarok the same way I implemented it in the bug
> fix<https://git.reviewboard.kde.org/r/109256/>attempt. For actually
> showing them on Amarok I’ll use the BookmarkGroupPtr
> to create a new bookmark group and add chapter pointers to it in the form
> of URLs. I would like to also store the chapter information in the database
> as well.
> 
> *2. Show MP3 chapter markers in Amarok*
> 
> Thanks to a recent pull request
> <https://github.com/taglib/taglib/pull/173>by krejclu6, MP3 chapter
> support has been implemented in TagLib. This small
> part of the project is to use this framework to show MP3 chapter markers in
> Amarok. The implementation of this in Amarok will be same as mentioned
> before.
> 
> *3. Implement OGG chapter support*
> 
> OGG chapter support is something TagLib still lacks. According to XIPH's
> docs <http://wiki.xiph.org/Chapter_Extension>, chapter support in OGG is
> very straightforward. The chapter data is found in the comments of the file
> along with the metadata.
> 
> *4. **Implement CUE Sheet support*
> 
> The fourth part of the project would be to implement complete CUE sheet
> support.
> 
> *What is CUE sheet?*
> 
> CUE sheet contains information related to the album, the main filename and
> the time offsets at which each track occurs.
> 
> *What needs to be done?*
> 
> I propose to make this change in the KUrl. Parts of the Url would have data
> about the Base File, Time-Offset and Duration of the track (after a “?” or
> “#” at the end, if not possible a new protocol?). Then these URL’s would be
> interpreted in the UmsCollection or SqlCollection classes as separate Track
> files. This would also enable storing the details of each song in the
> database, hence there is no need to access the CUE file every time.
> 
> Amarok already has a basic CUE sheet support, it can read CUE sheets added
> manually through the browser or from a CDROM. I would like to integrate
> this better with the CollectionScanner ( the way it is supposed to work,
> highlighted in the bug
> <https://bugs.kde.org/show_bug.cgi?id=187587>report's comments ). The
> scanner will automatically recognize these CUE
> files and record them as separate tracks in the database as mentioned
> above. Any changed metadata in the file would be stored in the KUrl.
> 
> *5. Support for NeroChapter type in MP4*
> 
> The last part of the project would be to implement support for the
> NeroChapter type in MP4. Since, this is not so popular, I would like to
> keep this the last part of the project. This involves more changes in
> TagLib than Amarok. The chapter track information is found under the “udta”
> atom. This part of the project would aim at automatically detecting the
> type of chapter present in the given file, decoding the chapter data and
> providing it neatly.
> 
> Tentative Timeline:
> 
> June
> < GSoC begins >
> 
> week 4: Community Bonding Period - Ask for improvement suggestions and get
> to know the Amarok community.
> 
> July-
> 
> week 1: Create Abstract Classes in TagLib for MP4 chapter support ( This
> will be useful for the last part of the project as well ). Implementing
> QtChapter (QuickTime) Support.
> 
> week 2: Finish implementing QtChapter (QuickTime) Support in TagLib.
> Implement OGG chapter support.
> 
> week 3: Modify the database and store the chapter data in it.
> 
> week 4: Tweak UI in Amarok to display the chapter markers (MP3, MP4,
> OGG). Start Implementation of CUE sheet support.
> 
> < Mid-term  - Complete chapter support in Amarok>
> 
> August-
> 
> *week 1*: Integrate CUE sheet reading with the scanner and automatically
> detect that these are separate track files and add them to the collection.
> 
> *weeks 2 and 3*: Implement changes in KUrl. Make modifications in
> SqlCollection and other methods and classes that use url(TrackPtr ,
> playback and other controllers).
> 
> *week 4*: Make modifications in UmsCollection to support CUE sheets as well.
> 
> September-
> 
> *week 1*: Implement modules that allow the user to change the metadata of
> CUE-sheeted tracks. I do not plan on making these changes in the CUE sheet
> instead store them locally/in the url so that Amarok can display them
> 
> *week 2*: Finish CUE sheet support in Amarok.
> 
> *week 3*: Implement NeroChapter support in TagLib.
> 
> < pencils down >
> 
> *week 4*: Testing and resolving any bugs.
> 
> Do you have other obligations from late May to early August (school, work,
> vacation, etc.)?
> 
> No. I can work upto 50 hours a week, 6-8 hours in the weekdays, and in the
> weekends too, if required, or if I am possessed with the compulsive urge to
> get something done ;) College starts early August, but it’s early in the
> semester, and the course work would be extremely light.
> 
> About Me:
> 
> I am an Information Technology undergrad at SSN College of Engineering. I
> am comfortable writing code in C/C++/Python. I was introduced to Amarok
> when my friend suggested it after looking at my android app
> Lyricize 
<https://play.google.com/store/apps/details?id=com.festalize.LyricMatch>.
> It totally drew me in. I'd love to make any contribution that makes
> music more enjoyable. I am also an enthusiastic web and android developer,
> and a great fan of FOSS. On a Foreign Exchange trade related project I
> recently worked on, I built a scraper
> <https://github.com/makkarlabs/stockscrape>, which I open sourced, that
> scrapes data from the National Stock Exchange website, and publishes it as
> a JSON API. I’ve also contributed to the Haiku OS.
> 
> Introduced to C and C++ at the age of 8, I continued to work on programming
> challenges, web and mobile related projects during school and college to
> sharpen my skills. I have learnt to code in languages such as java, php,
> ruby, node(javascript), etc. In the hackathon conducted by KRDS in 2012, I
> created a location guessing Facebook game which won the second
> prize<http://lighthouseinsights.in/a-chat-with-the-krds-india-facecode-hacka
> thon-winners.html>. I also was selected as one of the winners of Firefox OS
> hackathon conducted by Mozilla for creating a collaborative browser across
> platforms called BrowseWithMe using nodejs (
> server <https://bitbucket.org/abhiin1947/bb_monkey/src>,
> client <https://github.com/makkarlabs/bbmonkey> ). I created a
> gesturecontrolled web app 
> Fliptube <https://github.com/makkarlabs/fliptube>, a YouTube client where
> mobile users could flip through videos with ease, for the Yahoo! Open hack
> 2012 and it was selected as top 10 hacks of the night (out of 180+ hacks).
> 
> Am I comfortable working independently under a mentor or supervisor?
> 
> Yes. I have been an intern for the ULaw Software Foundation based in Canada
> and time zones aren’t really a problem at all.
> 
> After GSoC:
> 
> After GSoC I plan on continuing to fix bugs and code more features for
> Amarok and Taglib. I intend to learn about more multimedia related file
> formats and improving support in those areas.
> 
> One area I intentionally did not mention is the implementation of matroska
> chapter support. I plan on implementing this after GSoC as I feel it would
> be too much for 3 months. It is similar to the chapter markers in OGG
> Vorbis.
> Another requested feature is the support for embedded CUE files. I have
> explained these more clearly in the "Additional Information" section. This
> is one feature that requires a lot more time so, I plan to work on this
> after GSoC.
> 
> My experience with Amarok:
> 
> I started out my Amarok journey by trying to solve bug 240409 (which turns
> out to be a big part of my proposal ). It was initially marked as a junior
> job. I tried to implement chapter detection using library mp4v2 ( patch:
> https://git.reviewboard.kde.org/r/109079/ ) which was eventually rejected.
> After research and learning, I implemented support for chapter detection in
> Amarok by using mp4atom.h from taglib explicitly ( patch:
> https://git.reviewboard.kde.org/r/109256/ ). It was rejected because it was
> a technical preview. I have been busy with exams since, and have not been
> able to finish it.
> 
> Two other recent bug fixes that are not accepted yet -
> https://git.reviewboard.kde.org/r/110243/ ,
> https://git.reviewboard.kde.org/r/110150/
> 
> Additional Information:
> 
> Embedded CUE sheets are CUEs that are stored within the audio files. There
> are 2 formats in which they occur, FLAC and WAVPACK. The FLAC embedded CUE
> sheet has its own tags whereas the WAVPACK uses the apev2 tag. The task
> would be to implement CUE detection in TagLib and then use them to
> implement it in Amarok.
> 
> External Links:
> 
> (1) - https://bugs.kde.org/show_bug.cgi?id=187587 reported 2009
> 
> (2) - https://bugs.kde.org/show_bug.cgi?id=187587 ,
> https://bugs.kde.org/show_bug.cgi?id=286089 ,
> https://bugs.kde.org/show_bug.cgi?id=294136 ,
> https://bugs.kde.org/show_bug.cgi?id=270003 ,
> https://bugs.kde.org/show_bug.cgi?id=279234 ,
> https://bugs.kde.org/show_bug.cgi?id=230272
> 
> (3) - https://bugs.kde.org/show_bug.cgi?id=240409 ,
> https://bugs.kde.org/show_bug.cgi?id=187587 ,
> https://bugs.kde.org/show_bug.cgi?id=286089 ,
> https://bugs.kde.org/show_bug.cgi?id=270003 ,
> https://bugs.kde.org/show_bug.cgi?id=294136 ,
> https://bugs.kde.org/show_bug.cgi?id=144849
> 
> (4) - *https://bugs.kde.org/show_bug.cgi?id=144849*