Add torchscript export

src/python/larynx_train/export_torchscript.py

@@ -0,0 +1,63 @@
+#!/usr/bin/env python3
+import argparse
+import logging
+from pathlib import Path
+from typing import Optional
+
+import torch
+
+from .vits.lightning import VitsModel
+
+_LOGGER = logging.getLogger("larynx_train.export_torchscript")
+
+
+def main():
+    """Main entry point"""
+    torch.manual_seed(1234)
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint", help="Path to model checkpoint (.ckpt)")
+    parser.add_argument("output", help="Path to output model (.onnx)")
+
+    parser.add_argument(
+        "--debug", action="store_true", help="Print DEBUG messages to the console"
+    )
+    args = parser.parse_args()
+
+    if args.debug:
+        logging.basicConfig(level=logging.DEBUG)
+    else:
+        logging.basicConfig(level=logging.INFO)
+
+    _LOGGER.debug(args)
+
+    # -------------------------------------------------------------------------
+
+    args.checkpoint = Path(args.checkpoint)
+    args.output = Path(args.output)
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+
+    model = VitsModel.load_from_checkpoint(args.checkpoint)
+    model_g = model.model_g
+
+    num_symbols = model_g.n_vocab
+    num_speakers = model_g.n_speakers
+
+    # Inference only
+    model_g.eval()
+
+    with torch.no_grad():
+        model_g.dec.remove_weight_norm()
+
+    model_g.forward = model_g.infer
+
+    jitted_model = torch.jit.script(model_g)
+    torch.jit.save(jitted_model, str(args.output))
+
+    _LOGGER.info("Saved TorchScript model to %s", args.output)
+
+
+# -----------------------------------------------------------------------------
+
+if __name__ == "__main__":
+    main()