Text-to-Speech
ONNX
Safetensors
English
Chinese

Create model.json

#2
by pkufool - opened
Files changed (1) hide show
  1. model.json +29 -0
model.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model" : {
3
+ "fm-decoder-downsampling-factor" : "1,2,4,2,1",
4
+ "fm-decoder-num-layers" : "2,2,4,4,4",
5
+ "fm-decoder-cnn-module-kernel" : "31,15,7,15,31",
6
+ "fm-decoder-feedforward-dim" : 1536,
7
+ "fm-decoder-num-heads" : 4,
8
+ "fm-decoder-dim" : 512,
9
+ "text-encoder-downsampling-factor" : "1",
10
+ "text-encoder-num-layers" : "4",
11
+ "text-encoder-feedforward-dim" : 512,
12
+ "text-encoder-cnn-module-kernel" : 9,
13
+ "text-encoder-num-heads" : 4,
14
+ "text-encoder-dim" : 192,
15
+ "query-head-dim" : 32,
16
+ "value-head-dim" : 32,
17
+ "pos-head-dim" : 4,
18
+ "pos-dim" : 48,
19
+ "time-embed-dim" : 192,
20
+ "text-embed-dim" : 192
21
+ },
22
+ "feature" : {
23
+ "sampling_rate": 24000,
24
+ "frame_shift_ms": 256 / 24000 * 1000,
25
+ "feat_dim": 100,
26
+ "n_fft" : 1024,
27
+ "hop_length" : 256
28
+ }
29
+ }