update sherpa-onnx to use piper-phonemize

app.py

@@ -25,6 +25,7 @@ import uuid
 
 import gradio as gr
 import soundfile as sf
+import system
 
 from model import get_pretrained_model, language_to_models
 
@@ -185,7 +186,19 @@ with demo:
 
     gr.Markdown(description)
 
+
+def download_espeak_ng_data():
+    os.sytem(
+        """
+    cd /tmp
+    wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/espeak-ng-data.tar.bz2
+    tar xf espeak-ng-data.tar.bz2
+    """
+    )
+
+
 if __name__ == "__main__":
+    download_espeak_ng_data()
     formatter = "%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s"
 
     logging.basicConfig(format=formatter, level=logging.INFO)

model.py

@@ -124,12 +124,6 @@ def _get_vits_piper(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
         subfolder=".",
     )
 
-    lexicon = get_file(
-        repo_id=repo_id,
-        filename="lexicon.txt",
-        subfolder=".",
-    )
-
     tokens = get_file(
         repo_id=repo_id,
         filename="tokens.txt",
@@ -140,7 +134,8 @@ def _get_vits_piper(repo_id: str, speed: float) -> sherpa_onnx.OfflineTts:
         model=sherpa_onnx.OfflineTtsModelConfig(
             vits=sherpa_onnx.OfflineTtsVitsModelConfig(
                 model=model,
-                lexicon=lexicon,
+                lexicon="",
+                data_dir="/tmp/espeak-ng-data",
                 tokens=tokens,
                 length_scale=1.0 / speed,
             ),

requirements.txt

@@ -1,4 +1,4 @@
-https://huggingface.co/csukuangfj/wheels/resolve/main/sherpa_onnx-1.8.9-cp38-cp38-linux_x86_64.whl
+https://huggingface.co/csukuangfj/wheels/resolve/main/sherpa_onnx-1.8.11-cp38-cp38-linux_x86_64.whl
 #sherpa-onnx
 
 soundfile