Difference between revisions of "2014:Audio Fingerprinting"

From MIREX Wiki
m
Line 4: Line 4:
 
== Data ==
 
== Data ==
 
=== Database ===
 
=== Database ===
* 589 songs (*.mp3 and *.wav) corresponding to 573 different songs
+
* 10,000 songs (*.wav)
*Language: English, Chinese, Japanese, ...
+
* mono, 44.1 kHz, 16 bit resolution
* 35 mono, 554 stereo
 
* 40.97 hours long
 
  
 
=== Query ===
 
=== Query ===
* 305 recordings (*.wav) corresponding to 304 different songs in DB
+
* 1,264 10-second clips
* 166 mono, 139 stereo
+
* mono, 44.1 kHz, 16 bit resolution
* 16.11 hours long
 
 
* Recorded by variety brand of smartphones, containing noise
 
* Recorded by variety brand of smartphones, containing noise
* Variety of sampling rate and bit resolution
 
* Divided into 5800 10-second clips
 
  
 
== Evaluation Procedures ==
 
== Evaluation Procedures ==
Top-10 hit rate (1 point is scored for a hit in the top 10 and 0 is scored otherwise, as in the Query by Singing/Humming task)
+
Top-1 hit rate
  
 
== Submission Format ==
 
== Submission Format ==
 
Participants are required to submit a breakdown version of algorithm. The two parts are:
 
Participants are required to submit a breakdown version of algorithm. The two parts are:
  
1. Fingerprinting extractor
+
1. Database Builder
 
Command format:
 
Command format:
  extractor %file.list% %dir % %database_or_query%
+
  builder %file.list% %dir % %database_or_query%
 
where %file.list% is the input list of database audio files named as uniq_key.mp3. or uniq_key.wav For example:
 
where %file.list% is the input list of database audio files named as uniq_key.mp3. or uniq_key.wav For example:
 
  ./AFP/database/00001.wav
 
  ./AFP/database/00001.wav

Revision as of 07:34, 18 July 2014

Description

This task requires the query by using exact but noisy recordings.

Data

Database

  • 10,000 songs (*.wav)
  • mono, 44.1 kHz, 16 bit resolution

Query

  • 1,264 10-second clips
  • mono, 44.1 kHz, 16 bit resolution
  • Recorded by variety brand of smartphones, containing noise

Evaluation Procedures

Top-1 hit rate

Submission Format

Participants are required to submit a breakdown version of algorithm. The two parts are:

1. Database Builder Command format:

builder %file.list% %dir % %database_or_query%

where %file.list% is the input list of database audio files named as uniq_key.mp3. or uniq_key.wav For example:

./AFP/database/00001.wav
./AFP/database/00002.wav
./AFP/database/00003.mp3
./AFP/database/00004.wav
...

Output file(s) should be placed into %dir %

%database_or_query% is a string variable with two possible values. The value "DB" indicates the extractor should extract audio fingerprinting for the database (and builds index files, if it is needed); where the value "QUERY" indicates the extractor should extract audio fingerprinting for the query clips.

There are no limitations for output filename(s) for the database. However, your program should output one file for one query clip, and the main file name should be the same as given in the input %file.list%. We guarantee unique main filenames for database songs and query clips.

2. Matcher Command format:

matcher %dir_db% %dir_query% %resultFile%

where %dir_db% and %dir_query% are the directories which store output fingerprinting files (and possibly index files for database)

The result file gives top-10 candidates (if has) for each query. The format should be:

%main_query_file_name% %main_top_1_candiate_file_name% %main_top_2_candiate_file_name% …

For example:

q0001 0204 0048 9023 …
q0002 0043 8964 2378 …
q0003 2526 6782 3648 …
...

Please note that the order of output should be the alphabetic order of %main_query_file_name%.

Time and hardware limits

Due to the potentially high number of participants in this and other audio tasks, hard limits on the runtime of submissions are specified. The time/storage limits of different steps are shown in the following table:

Steps Time limit Storage (hard disk) limit
extractor (for database) Totally 72 hours 40KB for every 10 second audio + 250 MB extra overhead (if needed)
extractor (for query) 40 KB for each query clip
matcher N/A

Submissions that exceed these limitations may not receive a result.

Potential Participants

Discussion

name / email

Bibliography