Update README.md
Browse files
README.md
CHANGED
@@ -13,6 +13,17 @@ Try it with [ZFTurbo's Music-Source-Separation-Training](https://github.com/ZFTu
|
|
13 |
|
14 |
### Model
|
15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
Configs_256_8_4: [config_dereverb-echo_mel_band_roformer.yaml](./config_dereverb-echo_mel_band_roformer.yaml)<br>
|
17 |
Model_256_8_4: [dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt](./dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt)<br>
|
18 |
Instr dry sdr: 13.1507, Instr other sdr: 6.8830, Metric avg sdr: 10.0169
|
@@ -31,8 +42,6 @@ Datasets:
|
|
31 |
### Training log
|
32 |
|
33 |
Training logs: [train.log](./train.log)<br>
|
34 |
-
The following image is the TensorBoard visualization training log generated by [this script](./scripts/start_tensorboard.py).
|
35 |
-
![image](./tensorboard.png)
|
36 |
|
37 |
### Thanks
|
38 |
|
|
|
13 |
|
14 |
### Model
|
15 |
|
16 |
+
### V2 Models
|
17 |
+
|
18 |
+
Finetuned from: `dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt`<br>
|
19 |
+
Used 1000+ songs to Finetune.
|
20 |
+
|
21 |
+
Config: [config_dereverb_echo_mbr_v2.yaml](./config_dereverb_echo_mbr_v2.yaml)<br>
|
22 |
+
Model: [dereverb_echo_mbr_v2_sdr_dry_13.4843.ckpt](./dereverb_echo_mbr_v2_sdr_dry_13.4843.ckpt)<br>
|
23 |
+
Instr dry sdr: 13.4843 (Std: 4.8675)
|
24 |
+
|
25 |
+
#### V1 Models
|
26 |
+
|
27 |
Configs_256_8_4: [config_dereverb-echo_mel_band_roformer.yaml](./config_dereverb-echo_mel_band_roformer.yaml)<br>
|
28 |
Model_256_8_4: [dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt](./dereverb-echo_mel_band_roformer_sdr_10.0169.ckpt)<br>
|
29 |
Instr dry sdr: 13.1507, Instr other sdr: 6.8830, Metric avg sdr: 10.0169
|
|
|
42 |
### Training log
|
43 |
|
44 |
Training logs: [train.log](./train.log)<br>
|
|
|
|
|
45 |
|
46 |
### Thanks
|
47 |
|