2014:Multiple Fundamental Frequency Estimation & Tracking Results

Introduction

These are the results for the 2008 running of the Multiple Fundamental Frequency Estimation and Tracking task. For background information about this task set please refer to the 2013:Multiple Fundamental Frequency Estimation & Tracking page.

General Legend

Sub code	Submission name	Abstract	Contributors
CDM1	CDM1	PDF	Tian Cheng, Simon Dixon, Matthias Mauch
CDM2	CDM2	PDF	Tian Cheng, Simon Dixon, Matthias Mauch
CDM3	CDM3	PDF	Tian Cheng, Simon Dixon, Matthias Mauch
BW1	MSSIPLCA_fast_MultiF0	PDF	Emmanouil Benetos, Tillman Weyde
BW2	MSSIPLCA_fast_NoteTracking1	PDF	Emmanouil Benetos, Tillman Weyde
BW3	MSSIPLCA_fast_NoteTracking2	PDF	Emmanouil Benetos, Tillman Weyde

Task 1: Multiple Fundamental Frequency Estimation (MF0E)

MF0E Overall Summary Results

Below are the average scores across 40 test files. These files come from 3 different sources: woodwind quintet recording of bassoon, clarinet, horn,flute and oboe (UIUC); Rendered MIDI using RWC database donated by IRCAM and a quartet recording of bassoon, clarinet, violin and sax donated by Dr. Bryan Pardo`s Interactive Audio Lab (IAL). 20 files coming from 5 sections of the woodwind recording where each section has 4 files ranging from 2 polyphony to 5 polyphony. 12 files from IAL, coming from 4 different songs ranging from 2 polyphony to 4 polyphony and 8 files from RWC synthesized midi ranging from 2 different songs ranging from 2 polphony to 5 polyphony.

	BW1	CDM1	CDM2
Accuracy	0.662	0.620	0.620
Accuracy Chroma	0.694	0.660	0.660

download these results as csv

Detailed Results

	Precision	Recall	Accuracy	Etot	Esubs	Emiss	Efa
BW1	0.771	0.735	0.662	0.392	0.112	0.153	0.128
CDM1	0.741	0.694	0.620	0.443	0.128	0.178	0.137
CDM2	0.741	0.694	0.620	0.443	0.128	0.178	0.137

download these results as csv

Detailed Chroma Results

Here, accuracy is assessed on chroma results (i.e. all F0's are mapped to a single octave before evaluating)

	Precision	Recall	Accuracy	Etot	Esubs	Emiss	Efa
BW1	0.808	0.774	0.694	0.354	0.073	0.153	0.128
CDM1	0.789	0.742	0.660	0.395	0.080	0.178	0.137
CDM2	0.789	0.742	0.660	0.395	0.080	0.178	0.137

download these results as csv

Individual Results Files for Task 1

BW1= Emmanouil Benetos, Tillman Weyde
CDM1= Tian Cheng, Simon Dixon, Matthias Mauch
CDM2= Tian Cheng, Simon Dixon, Matthias Mauch

Info about the filenames

The filenames starting with part* comes from acoustic woodwind recording, the ones starting with RWC are synthesized. The legend about the instruments are:

bs = bassoon, cl = clarinet, fl = flute, hn = horn, ob = oboe, vl = violin, cel = cello, gtr = guitar, sax = saxophone, bass = electric bass guitar

Run Times

sub_ID	Runtime(sec)
BW1	3374
CDM1	62027
CDM2	61992

download these results as csv

Friedman tests for Multiple Fundamental Frequency Estimation (MF0E)

The Friedman test was run in MATLAB to test significant differences amongst systems with regard to the performance (accuracy) on individual files.

Tukey-Kramer HSD Multi-Comparison

TeamID	TeamID	Lowerbound	Mean	Upperbound	Significance
BW1	CDM2	0.8211	1.2750	1.7289	TRUE
BW1	CDM1	0.8211	1.2750	1.7289	TRUE
CDM2	CDM1	-0.4539	0.0000	0.4539	FALSE

download these results as csv

Task 2:Note Tracking (NT)

NT Mixed Set Overall Summary Results

This subtask is evaluated in two different ways. In the first setup , a returned note is assumed correct if its onset is within +-50ms of a ref note and its F0 is within +- quarter tone of the corresponding reference note, ignoring the returned offset values. In the second setup, on top of the above requirements, a correct returned note is required to have an offset value within 20% of the ref notes duration around the ref note`s offset, or within 50ms whichever is larger.

A total of 34 files were used in this subtask: 16 from woodwind recording, 8 from IAL quintet recording and 6 piano.

	BW2	BW3	CDM3
Ave. F-Measure Onset-Offset	0.3264	0.2745	0.2880
Ave. F-Measure Onset Only	0.5533	0.4754	0.5071
Ave. F-Measure Chroma	0.3383	0.3142	0.3058
Ave. F-Measure Onset Only Chroma	0.5751	0.5512	0.5439

download these results as csv

Detailed Results

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.335	0.334	0.326	0.879
BW3	0.237	0.342	0.274	0.879
CDM3	0.268	0.328	0.288	0.857

download these results as csv

Detailed Chroma Results

Here, accuracy is assessed on chroma results (i.e. all F0's are mapped to a single octave before evaluating)

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.347	0.347	0.338	0.875
BW3	0.270	0.395	0.314	0.877
CDM3	0.283	0.350	0.306	0.854

download these results as csv

Results Based on Onset Only

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.592	0.546	0.553	0.726
BW3	0.419	0.576	0.475	0.720
CDM3	0.479	0.562	0.507	0.702

download these results as csv

Chroma Results Based on Onset Only

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.615	0.570	0.575	0.712
BW3	0.483	0.675	0.551	0.684
CDM3	0.513	0.606	0.544	0.682

download these results as csv

Run Times

sub_ID	Runtime(sec)
BW2	3706
BW3	2203
CDM3	68470

download these results as csv

Friedman Tests for Note Tracking

The Friedman test was run in MATLAB to test significant differences amongst systems with regard to the F-measure on individual files.

Tukey-Kramer HSD Multi-Comparison for Task2

TeamID	TeamID	Lowerbound	Mean	Upperbound	Significance
BW2	CDM3	0.0786	0.6471	1.2155	TRUE
BW2	BW3	0.2845	0.8529	1.4214	TRUE
CDM3	BW3	-0.3625	0.2059	0.7743	FALSE

download these results as csv

NT Piano-Only Overall Summary Results

This subtask is evaluated in two different ways. In the first setup , a returned note is assumed correct if its onset is within +-50ms of a ref note and its F0 is within +- quarter tone of the corresponding reference note, ignoring the returned offset values. In the second setup, on top of the above requirements, a correct returned note is required to have an offset value within 20% of the ref notes duration around the ref note`s offset, or within 50ms whichever is larger. 6 piano recordings are evaluated separately for this subtask.

	BW2	BW3	CDM3
Ave. F-Measure Onset-Offset	0.1596	0.1940	0.1551
Ave. F-Measure Onset Only	0.5359	0.6110	0.5723
Ave. F-Measure Chroma	0.1653	0.2055	0.1692
Ave. F-Measure Onset Only Chroma	0.5455	0.6274	0.5904

download these results as csv

Detailed Results

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.186	0.141	0.160	0.831
BW3	0.193	0.195	0.194	0.831
CDM3	0.158	0.153	0.155	0.788

download these results as csv

Detailed Chroma Results

Here, accuracy is assessed on chroma results (i.e. all F0's are mapped to a single octave before evaluating)

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.192	0.146	0.165	0.824
BW3	0.205	0.207	0.205	0.819
CDM3	0.172	0.167	0.169	0.780

download these results as csv

Results Based on Onset Only

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.610	0.483	0.536	0.559
BW3	0.608	0.616	0.611	0.562
CDM3	0.575	0.574	0.572	0.538

download these results as csv

Chroma Results Based on Onset Only

	Precision	Recall	Ave. F-measure	Ave. Overlap
BW2	0.622	0.491	0.546	0.555
BW3	0.624	0.633	0.627	0.557
CDM3	0.593	0.592	0.590	0.538

download these results as csv

Individual Results Files for Task 2

BW2= Emmanouil Benetos, Tillman Weyde
BW3= Emmanouil Benetos, Tillman Weyde
CDM3= Tian Cheng, Simon Dixon, Matthias Mauch