Upload 2 files

config/inference/MossFormer2_SR_48K.json

@@ -0,0 +1,32 @@
+{
+    "resblock": "1",
+    "num_gpus": 1,
+    "batch_size": 2,
+    "learning_rate": 0.0001,
+    "learning_rate_mossformer": 0.0005,
+    "adam_b1": 0.8,
+    "adam_b2": 0.99,
+    "lr_decay": 0.999,
+    "seed": 1234,
+    "upsample_rates": [8, 8, 2, 2],
+    "upsample_kernel_sizes": [16, 16, 4, 4],
+    "upsample_initial_channel": 1024,
+    "resblock_kernel_sizes": [3, 7, 11],
+    "resblock_dilation_sizes": [[1, 3, 5], [1, 3, 5], [1, 3, 5]],
+    "segment_size": 24576,
+    "num_mels": 80,
+    "num_freq": 1025,
+    "n_fft": 1024,
+    "hop_size": 256,
+    "win_size": 1024,
+    "sampling_rate": 48000,
+    "fmin": 0,
+    "fmax": 8000,
+    "fmax_for_loss": null,
+    "num_workers": 4,
+    "dist_config": {
+        "dist_backend": "nccl",
+        "dist_url": "tcp://localhost:54321",
+        "world_size": 1
+    }
+}

config/inference/MossFormer2_SR_48K.yaml

@@ -0,0 +1,16 @@
+#!/bin/bash 
+mode: 'inference'
+config_json: "config/inference/MossFormer2_SR_48K.json"
+use_cuda: 1 # 1 for True, 0 for False
+num_gpu: 1
+sampling_rate: 48000
+network: "MossFormer2_SR_48K"  ##network type
+checkpoint_dir: "checkpoints/MossFormer2_SR_48K"
+
+#input_path: support wav dir or wav scp or a wav file
+input_path: "samples/scp/cv_demand_testset_16k.scp"
+output_dir: "outputs/MossFormer2_SR_48K"
+
+# decode parameters
+one_time_decode_length: 20 #maximum segment length for one-pass decoding (seconds), longer audio  will use segmented decoding
+decode_window: 4 #one-pass decoding length