groadabike
commited on
Commit
•
96a3d10
1
Parent(s):
697547a
Update README.md
Browse files
README.md
CHANGED
@@ -1,9 +1,111 @@
|
|
1 |
---
|
|
|
|
|
|
|
2 |
tags:
|
3 |
-
-
|
4 |
-
-
|
|
|
|
|
|
|
|
|
|
|
5 |
---
|
6 |
|
7 |
-
|
8 |
-
|
9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
license: apache-2.0
|
3 |
+
language:
|
4 |
+
- en
|
5 |
tags:
|
6 |
+
- hearing loss
|
7 |
+
- challenge
|
8 |
+
- signal processing
|
9 |
+
- source separation
|
10 |
+
- lyrics intelligibility
|
11 |
+
- audio
|
12 |
+
- audio-to-audio
|
13 |
---
|
14 |
|
15 |
+
# Cadenza Challenge: CAD2-Task1
|
16 |
+
|
17 |
+
A NonCausal Lyrics/Accompaniment separation model for the CAD2-Task1 baseline system.
|
18 |
+
|
19 |
+
* Architecture: ConvTasNet (Kaituo XU) with multichannel support (Alexandre Defossez).
|
20 |
+
* Parameters:
|
21 |
+
* B: 256
|
22 |
+
* C: 2
|
23 |
+
* H: 512
|
24 |
+
* L: 20
|
25 |
+
* N: 256
|
26 |
+
* P: 3
|
27 |
+
* R: 4
|
28 |
+
* X: 10
|
29 |
+
* audio_channels: 2
|
30 |
+
* causal: false
|
31 |
+
* mask_nonlinear: relu
|
32 |
+
* norm_type: gLN
|
33 |
+
* training:
|
34 |
+
* sample_rate: 44100
|
35 |
+
* samples_per_track: 64
|
36 |
+
* segment: 5.0
|
37 |
+
* aggregate: 2
|
38 |
+
* batch_size: 4
|
39 |
+
* early_stop: true
|
40 |
+
* epochs: 200
|
41 |
+
|
42 |
+
|
43 |
+
## Dataset
|
44 |
+
The model was trained on the training split of the MUSDB18-HQ dataset.
|
45 |
+
|
46 |
+
## How to use
|
47 |
+
|
48 |
+
```
|
49 |
+
from tasnet import ConvTasNetStereo
|
50 |
+
|
51 |
+
model = ConvTasNetStereo.from_pretrained(
|
52 |
+
"cadenzachallenge/ConvTasNet_LyricsSeparation_NonCausal"
|
53 |
+
).cpu()
|
54 |
+
|
55 |
+
```
|
56 |
+
|
57 |
+
## Results
|
58 |
+
|
59 |
+
| Track | Vocals (SDR) | Accompaniment (SDR) |
|
60 |
+
|:------|:------------:|:---------:|
|
61 |
+
| Al James - Schoolboy Facination | 6.841 | 9.074 |
|
62 |
+
| AM Contra - Heart Peripheral | 6.948 | 14.105 |
|
63 |
+
| Angels In Amplifiers - I'm Alright | 7.358 | 10.859 |
|
64 |
+
| Arise - Run Run Run | 6.105 | 16.806 |
|
65 |
+
| Ben Carrigan - We'll Talk About It All Tonight | 2.853 | 10.181 |
|
66 |
+
| BKS - Bulldozer | 1.909 | 13.944 |
|
67 |
+
| BKS - Too Much | 8.615 | 13.212 |
|
68 |
+
| Bobby Nobody - Stitch Up | 7.948 | 12.685 |
|
69 |
+
| Buitraker - Revo X | 4.609 | 14.61 |
|
70 |
+
| Carlos Gonzalez - A Place For Us | 4.235 | 8.888 |
|
71 |
+
| Cristina Vane - So Easy | 8.759 | 13.639 |
|
72 |
+
| Detsky Sad - Walkie Talkie | 7.732 | 10.844 |
|
73 |
+
| Enda Reilly - Cur An Long Ag Seol | 9.603 | 13.723 |
|
74 |
+
| Forkupines - Semantics | 4.955 | 11.561 |
|
75 |
+
| Georgia Wonder - Siren | 4.124 | 8.578 |
|
76 |
+
| Girls Under Glass - We Feel Alright | 4.38 | 12.272 |
|
77 |
+
| Hollow Ground - Ill Fate | 7.046 | 16.299 |
|
78 |
+
| James Elder & Mark M Thompson - The English Actor | 4.694 | 9.638 |
|
79 |
+
| Juliet's Rescue - Heartbeats | 6.281 | 14.409 |
|
80 |
+
| Little Chicago's Finest - My Own | 6.313 | 6.603 |
|
81 |
+
| Louis Cressy Band - Good Time | 6.501 | 12.016 |
|
82 |
+
| Lyndsey Ollard - Catching Up | 9.18 | 12.116 |
|
83 |
+
| M.E.R.C. Music - Knockout | 6.619 | 8.507 |
|
84 |
+
| Moosmusic - Big Dummy Shake | 8.097 | 14.578 |
|
85 |
+
| Motor Tapes - Shore | 0.769 | 10.137 |
|
86 |
+
| Mu - Too Bright | 5.853 | 13.135 |
|
87 |
+
| Nerve 9 - Pray For The Rain | 6.425 | 14.427 |
|
88 |
+
| PR - Happy Daze | 0 | 51.092 |
|
89 |
+
| PR - Oh No | 0 | 9.021 |
|
90 |
+
| Punkdisco - Oral Hygiene | 5.725 | 17.681 |
|
91 |
+
| Raft Monk - Tiring | 2.378 | 9.244 |
|
92 |
+
| Sambasevam Shanmugam - Kaathaadi | 8.164 | 10.588 |
|
93 |
+
| Secretariat - Borderline | 5.522 | 10.817 |
|
94 |
+
| Secretariat - Over The Top | 7.859 | 14.996 |
|
95 |
+
| Side Effects Project - Sing With Me | 11.197 | 12.63 |
|
96 |
+
| Signe Jakobsen - What Have You Done To Me | 7.685 | 11.013 |
|
97 |
+
| Skelpolu - Resurrection | 0 | 7.603 |
|
98 |
+
| Speak Softly - Broken Man | 3.997 | 14.516 |
|
99 |
+
| Speak Softly - Like Horses | 6.462 | 9.426 |
|
100 |
+
| The Doppler Shift - Atrophy | 0.711 | 14.241 |
|
101 |
+
| The Easton Ellises - Falcon 69 | 2.401 | 7.889 |
|
102 |
+
| The Easton Ellises (Baumi) - SDRNR | 1.479 | 7.948 |
|
103 |
+
| The Long Wait - Dark Horses | 6.53 | 12.661 |
|
104 |
+
| The Mountaineering Club - Mallory | 10.665 | 15.311 |
|
105 |
+
| The Sunshine Garcia Band - For I Am The Moon | 9.591 | 13.297 |
|
106 |
+
| Timboz - Pony | 4.025 | 14.271 |
|
107 |
+
| Tom McKenzie - Directions | 8.031 | 16.129 |
|
108 |
+
| Triviul feat. The Fiend - Widow | 7.061 | 8.168 |
|
109 |
+
| We Fell From The Sky - Not You | 3.862 | 11.685 |
|
110 |
+
| Zeno - Signs | 6.364 | 11.552 |
|
111 |
+
| **Total (median over frames, median over tracks)** | **6.338** | **12.194** |
|