add onnx export

rvc/modules/onnx/export.py

@@ -0,0 +1,52 @@
+import torch
+
+from rvc.lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
+
+
+def export_onnx(ModelPath, ExportedPath):
+    cpt = torch.load(ModelPath, map_location="cpu")
+    cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]
+    vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768
+
+    test_phone = torch.rand(1, 200, vec_channels)  # hidden unit
+    test_phone_lengths = torch.tensor([200]).long()  # hidden unit length (doesn't seem to help)）
+    test_pitch = torch.randint(size=(1, 200), low=5, high=255)  # Base frequency (in Hz)
+    test_pitchf = torch.rand(1, 200)  # nsf base frequency
+    test_ds = torch.LongTensor([0])  # Speaker ID
+    test_rnd = torch.rand(1, 192, 200)  # Noise (add random factor)
+
+    device = "cpu"  # Device on export (does not affect use of model)
+
+    net_g = SynthesizerTrnMsNSFsidM(
+        *cpt["config"], is_half=False, version=cpt.get("version", "v1")
+    )  # fp32 export (C++ has to manually rearrange memory to support fp16 so no fp16 for now)
+    net_g.load_state_dict(cpt["weight"], strict=False)
+    input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"]
+    output_names = [
+        "audio",
+    ]
+    # net_g.construct_spkmixmap(n_speaker) Multi-Role Mixed Track Export
+    torch.onnx.export(
+        net_g,
+        (
+            test_phone.to(device),
+            test_phone_lengths.to(device),
+            test_pitch.to(device),
+            test_pitchf.to(device),
+            test_ds.to(device),
+            test_rnd.to(device),
+        ),
+        ExportedPath,
+        dynamic_axes={
+            "phone": [1],
+            "pitch": [1],
+            "pitchf": [1],
+            "rnd": [2],
+        },
+        do_constant_folding=False,
+        opset_version=13,
+        verbose=False,
+        input_names=input_names,
+        output_names=output_names,
+    )
+    return "Finished"