Difference between revisions of "2014:Singing Voice Separation"

From MIREX Wiki
m (Evaluation)
m (Evaluation)
Line 39: Line 39:
 
       4.4363
 
       4.4363
  
The final scores will be determined by the average scores of all 100 songs:
+
The final scores will be determined by the average scores over all 100 clips:
  
 
<math>GSDR=\frac{\sum_{i=1}^{100}SDR_i}{100}</math>,
 
<math>GSDR=\frac{\sum_{i=1}^{100}SDR_i}{100}</math>,

Revision as of 01:39, 2 September 2014

Description

The singing voice separation task solicits competing entries to blindly separate the singer's voice from pop music recordings. The entries are evaluated using standard metrics (see Evaluation below).

Task specific mailing list

All discussions take place on the MIREX "EvalFest" list. If you have an question or comment, simply include the task name in the subject heading.

Data

A collection of 100 clips of recorded pop music (vocals plus music) are used to evaluate the singing voice separation algorithms.

Collection statistics:

  1. Size of collection: 100 clips
  2. Audio details: 16-bit, mono, 44.1kHz, WAV
  3. Duration of each clip: 30 seconds

Evaluation

For evaluation we use Vincent et al.'s (2012) Source to Distortion Ratio (SDR), Source to Interferences Ratio (SIR), and Sources to Artifacts Ratio (SAR), as implemented by bss_eval_sources.m in BSS Eval Version 3.0. More specifically, their function will be invoked as follows:

>> trueVoice = wavread('trueVoice.wav');
>> trueKaraoke = wavread('trueKaraoke.wav');
>> trueMixed = trueVoice + trueKaraoke;
>> [estimatedVoice, estimatedKaraoke] = wrapper_function_calling_your_separation_algorithm(trueMixed);
>> [SDR, SIR, SAR] = bss_eval_sources([estimatedVoice estimatedKaraoke]', [trueVoice trueKaraoke]')

 SDR =
    -2.7443
     2.7514

 SIR =
    -0.5416
     9.0144

 SAR =
     4.5486
     4.4363

The final scores will be determined by the average scores over all 100 clips:

,

,

.

Submission format

Participants are required to submit an entry that takes in a filename (in the form of *.wav) as its only argument. The entries must send their voice-separated outputs to *-voice.wav and *-music.wav, respectively.

Packaging submissions

All submissions should be statically linked to all libraries (the presence of dynamically linked libraries cannot be guarenteed).

All submissions should include a README file including the following the information:

  1. Command line calling format for all executables and an example formatted set of commands
  2. Number of threads/cores used or whether this should be specified on the command line
  3. Expected memory footprint
  4. Expected runtime
  5. Any required environments (and versions), e.g. python, java, bash, matlab.

Time and hardware limits

Due to the potentially high number of particpants in this and other audio tasks, hard limits on the runtime of submissions are specified.

A hard limit of 24 hours will be imposed on runs. Submissions that exceed this runtime may not receive a result.

Potential Participants

name / email