Text to audio

model.py

@@ -1,4 +1,6 @@
 from functools import lru_cache
+import os
+import tempfile
 from pathlib import Path
 
 import sherpa_onnx
@@ -33,11 +35,20 @@ def get_pretrained_model(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
         filename="model.onnx",
         subfolder="kokoro-en-v0_19",
     )
-    tokens = _get_file(
+    tokens_raw = _get_file(
         repo_id=source_repo,
         filename="tokens.txt",
         subfolder="kokoro-en-v0_19",
     )
+
+    # Sanitize tokens file to prevent parsing errors (e.g. empty lines)
+    with open(tokens_raw, "r", encoding="utf-8") as f:
+        lines = [line for line in f if line.strip()]
+
+    fd, tokens = tempfile.mkstemp(suffix=".txt", text=True)
+    with os.fdopen(fd, "w", encoding="utf-8") as f:
+        f.writelines(lines)
+
     voices = _get_file(
         repo_id=source_repo,
         filename="voices.bin",
@@ -66,4 +77,3 @@ def get_pretrained_model(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
         max_num_sentences=1,
     )
     return sherpa_onnx.OfflineTts(tts_config)
-