Singer Diarization for Polyphonic Music with Unison Singing: Datasets Detail

Paper: Hitoshi Suda, Daisuke Saito, Satoru Fukayama, Tomoyasu Nakano, and Masataka Goto. Singer Diarization for Polyphonic Music With Unison Singing. IEEE/ACM Transactions on Audio, Speech, and Language Processing, Vol. 30, May 2022.

This web page describes detailed information about datasets that we used for experiments in the paper.

This page shows a title, duration, and a catalog number of the compact disc (CD) of each song.

In addition, the label files that we manually annotated on the songs are available in this page.

Because of the copyright issues, we are not allowed to share musical audio signals of the songs used in our experiments.

You can get the same data by buying the CDs with the shown catalog numbers.

For convenience, we have provided YouTube links by which people can listen to some songs only when they are available on YouTube, but please note that YouTube videos are often excerpts (short versions).

Label format

[start time] [end time] [singer ids]
[start time] [end time] [singer ids]
...

Example

29.127 42.616 1
42.616 55.625 2
56.190 66.269 3
66.269 79.989 1,2,3
79.989 82.891 1
82.891 83.756 3
83.756 84.628 2
84.628 93.548 1,2,3
93.548 96.653 2
96.653 97.473 1
97.473 98.290 3
98.290 100.223 1,2,3
108.081 121.580 2
123.334 127.970 3
127.970 130.076 1,2,3
130.076 136.128 3
136.774 147.043 1
147.043 161.436 1,2,3
161.436 164.370 3
164.370 165.217 1
165.217 166.082 2
166.082 167.972 1,2,3
195.875 198.951 2
198.951 202.728 1
202.728 205.878 3
208.328 208.759 1
208.759 209.163 3
209.163 209.645 2
209.645 212.731 3
212.731 216.113 2
216.113 219.253 1
219.253 229.929 1,2,3
229.929 232.895 3
232.895 233.848 1
233.848 234.663 2
234.663 255.692 1,2,3

Dataset A

Dataset A is used for experiments to evaluate Cosacorr scores (Section V in the paper).

Purpose	Title	Duration	Catalog number
Training	9:02 pm	2:03	COCX-34830–4
	エージェント夜を往く (Agent Yoru Wo Yuku)	2:04	COCX-34830–4
	蒼い鳥 (Aoi Tori)	2:06	COCX-34830–4
	おはよう!! 朝ご飯 (Ohayou!! Asagohan)	2:01	COCX-34830–4
	きゅんっ! ヴァンパイアガール (Kyun! Vampire Girl)	2:00	SACX-1022–3
	キラメキラリ (Kiramekirari)	2:02	COCX-36135–7
	Little Match Girl	2:08	SACX-1022–3
	迷走Mind (Meiso Mind)	2:04	COCX-36135–7
	relations	2:05	COCX-34830–4
	隣に… (Tonarini...)	2:04	COCX-36135–7
Evaluation (closed-singer)	Kosmos, Cosmos	2:01	COCX-36135–7
Evaluation (open-singer)	ハイファイ☆デイズ (Hi-fi Days)	3:49	SACX-1040

Dataset B

Dataset B is a training set described in Section VI. A. 1).

Purpose	Title	Duration	Catalog number
Training	エージェント夜を往く (Agent Yoru Wo Yuku)	2:04	COCX-34830–4
	蒼い鳥 (Aoi Tori)	2:06	COCX-34830–4
	おはよう!! 朝ご飯 (Ohayou!! Asagohan)	2:01	COCX-34830–4
	きゅんっ! ヴァンパイアガール (Kyun! Vampire Girl)	2:00	SACX-1022–3
	キラメキラリ (Kiramekirari)	2:02	COCX-36135–7
	Kosmos, Cosmos	2:01	COCX-36135–7
	Little Match Girl	2:08	SACX-1022–3
	迷走Mind (Meiso Mind)	2:04	COCX-36135–7
	relations	2:05	COCX-34830–4
	隣に… (Tonarini...)	2:04	COCX-36135–7
	フタリの記憶 (Futari No Kioku)	2:03	COCX-36135–7
	Honey Heartbeat	2:02	SACX-1022–3
	MEGARE!	2:03	SACX-1022–3
	スタ→トスタ→ (Start Star)	2:00	COCX-36135–7
	THE IDOLM@STER	2:03	COCX-34830–4
	だってだって噫無情 (Datte Datte Aa Mujo)	5:31	LACA-39600–26
	愛してるばんざーい! (Aishiteru Banzai!)	4:51	LACX-39331–9
	Angelic Angel	4:55	LACA-39600–26
	嵐のなかの恋だから (Arashi No Naka No Koi Dakara)	5:12	LACA-39600–26
	僕たちはひとつの光 (Bokutachi Wa Hitotsu No Hikari)	4:52	LACA-39600–26
	僕らは今のなかで (Bokura Wa Ima No Naka De)	4:34	LACX-39331–9
	COLORFUL VOICE	3:39	LACA-39600–26
	Dancing stars on me!	4:30	LACA-39600–26
	どんなときもずっと (Donna Toki Mo Zutto)	4:35	LACA-39600–26
	Happy maker!	4:54	LACA-39600–26
	HEART to HEART!	4:37	LACA-39600–26
	輝夜の城で踊りたい (Kaguya No Shiro De Odoritai)	4:31	LACX-39331–9
	KiRa-KiRa Sensation!	4:55	LACA-39600–26
	これからのSomeday (Korekara No Someday)	4:25	LACX-39331–9
	LOVELESS WORLD	5:13	LACA-39600–26
	Oh, Love & Peace!	4:58	LACX-39331–9
	Love wing bell	4:45	LACA-39600–26
	ミはμ'sicのミ (Mi Wa μ'sic No Mi)	4:35	LACA-39600–26
	もぎゅっと“love”で接近中! (Mogyutto “love” De Sekkin Chu!)	5:43	LACX-39331–9
	MOMENT RING	6:09	LACA-39600–26
	Music S.T.A.R.T!!	4:54	LACA-39600–26
	No brand girls	4:04	LACX-39331–9
	Paradise Live	5:08	LACA-39600–26
	るてしキスキしてる (Ruteshikisukishiteru)	5:29	LACA-39600–26
	さようならへさよなら! (Sayonara E Sayonara!)	5:06	LACA-39600–26
	きっと青春が聞こえる (Kitto Seishun Ga Kikoeru)	4:07	LACX-39331–9
	SENTIMENTAL StepS	3:57	LACA-39600–26
	Shangri-La Shower	5:09	LACA-39600–26
	それは僕たちの奇跡 (Sore Wa Bokutachi No Kiseki)	4:15	LACA-39600–26
	START:DASH!!	4:18	LACX-39331–9
	SUNNY DAY SONG	4:45	LACA-39600–26
	Super LOVE=Super LIVE!	5:38	LACA-39600–26
	タカラモノズ (Takaramonozu)	3:44	LACA-39600–26
	WILD STARS	4:19	LACX-39331–9
	Wonderful Rush	4:42	LACX-39331–9
	ユメノトビラ (Yume No Tobira)	4:48	LACA-39600–26
Development	9:02 pm	2:03	COCX-34830–4
Development	Wonder zone	3:57	LACX-39331–9

Dataset C

Dataset C is a training set for voice activity and overlap detection described in Section VI. A. 2).

Label data are available. These data include only states of voice activity and overlap.

Purpose	Title	Duration	Catalog number	Sample
Training	GO MY WAY!!	4:50	COCX-36239
	LOST	5:52	COCX-36239
	My Best Friend	4:21	COCX-36239
	いっしょ (Issho)	4:51	COCX-36239
	またね (Matane)	5:02	COCX-36239
	エージェント夜を往く (Agent Yoru Wo Yuku)	4:11	COCX-36239
	キミはメロディ	4:01	COCX-36239
	after school NAVIGATORS	4:48	LACA-39260–1
	Cutie Panther	4:30	LACA-9393–5
	Listen to my heart!!	4:28	LACA-39260–1
	Love marginal	4:35	LACA-39260–1
	Mermaid festa vol.2 ～Passionate～	5:25	LACA-39260–1
	Pure girls project	4:35	LACA-9393–5
	soldier game	3:42	LACA-39260–1
	START:DASH!!	4:17	LACA-9393–5	YouTube
	sweet & sweet holiday	4:17	LACA-39260–1
	UNBALANCED LOVE	4:26	LACA-9393–5
	これからのSomeday (Korekara No Someday)	4:26	LACA-9393–5
	微熱からMystery (Binetsu Kara Mystery)	3:19	LACA-9393–5
Development (closed-singer)	思い出をありがとう (Omoide Wo Arigato)	4:29	COCX-36239
Development (closed-singer)	告白日和、です! (Kokuhaku Biyori, Desu!)	3:38	LACA-39260–1
Development (open-singer)	NEO THEORY FANTASY	5:50	LACM-14863	YouTube
	ラビリンス・レジスタンス (Labyrinth Resistance)	3:37	LACM-14863	YouTube
	ステップ&スキップ (Step & Skip)	4:16	COCC-17702	YouTube
	幸せの法則～ルール～ (Shiawase No Rule)	4:46	COCC-17706

Dataset D

Dataset D is a evaluation set described in Section VI. A. 3).

Label data are available.

Title	Duration	Catalog number	Sample
アルストロメリア (Alstroemeria)	4:47	LACM-14785	YouTube
ハピリリ (Happy Lily)	4:23	LACM-14785	YouTube
ZETTAI×BREAK!! トゥインクルリズム (ZETTAI×BREAK!! Twinkle Rhythm)	4:27	LACM-14637	YouTube
Tomorrow Program	4:59	LACM-14637	YouTube
Melty Fantasia	4:23	LACM-14638	YouTube
I.D 〜EScape from Utopia〜	3:48	LACM-14638	YouTube
花ざかりWeekend✿ (Hanazakari Weekend✿)	4:12	LACM-14639	YouTube
RED ZONE	4:11	LACM-14639	YouTube
咲くは浮世の君花火 (Saku Wa Ukiyo No Kimi Hanabi)	4:21	LACM-14640	YouTube
BORN ON DREAM! 〜HANABI☆NIGHT〜	4:06	LACM-14640	YouTube
だってあなたはプリンセス (Datte Anata Wa Princess)	4:42	LACM-14824	YouTube
ミラージュ・ミラー (Mirage Mirror)	4:31	LACM-14824	YouTube
月曜日のクリームソーダ (Getsuyoubi No Cream Soda)	4:23	LACM-14825	YouTube
I did+I will	4:38	LACM-14825	YouTube
ピコピコIIKO! インベーダー (Pikopiko IIKO! Invader)	4:18	LACM-14826	YouTube
Get lol! Get lol! SONG	4:18	LACM-14826	YouTube
dans l'obscurité	5:24	LACM-14892	YouTube
囚われのTeaTime (Toraware No TeaTime)	5:33	LACM-14892	YouTube
Tulip	3:53	COCC-17142	YouTube
ハイファイ☆デイズ (Hi-fi Days)	3:49	COCC-17143	YouTube
イリュージョニスタ! (Illusionista!)	4:17	COCC-17458	YouTube
Yes! Party Time!!	4:21	COCC-17261	YouTube
shabon song	4:40	COCC-17153	YouTube
ガールズ・イン・ザ・フロンティア (Girls in the Frontier)	4:14	COCC-17520	YouTube
夢で夜空を照らしたい (Yume De Yozora Wo Terashitai)	5:34	LACM-14522	YouTube