Download vosk model in package

speech_recognition/cli.py

@@ -3,6 +3,7 @@
 import shutil
 import tempfile
 import zipfile
+from pathlib import Path
 from urllib.request import urlopen
 
 from tqdm import tqdm
@@ -51,10 +52,11 @@ def main(argv=None) -> None:
         "--url",
         default="https://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip",
     )
-    vosk_parser.add_argument("--dir", default="model")
 
     def _download_vosk(args):
-        download_vosk_model(args.url, args.dir)
+        download_vosk_model(
+            args.url, str(Path(__file__).parent / "models" / "vosk")
+        )
 
     vosk_parser.set_defaults(func=_download_vosk)
 

speech_recognition/models/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

speech_recognition/recognizers/vosk.py

@@ -1,9 +1,11 @@
 from __future__ import annotations
 
 import json
-import os
+from pathlib import Path
 from typing import TYPE_CHECKING, Literal, TypedDict, Union, cast, overload
 
+from speech_recognition.exceptions import SetupError
+
 if TYPE_CHECKING:
     from speech_recognition.audio import AudioData
 
@@ -38,11 +40,15 @@ def recognize(
 
     from vosk import KaldiRecognizer, Model
 
-    if not os.path.exists("model"):
-        return "Please download the model from https://alphacephei.com/vosk/models and unpack as 'model' in the current folder."
+    vosk_model_path = Path(__file__).parent.parent / "models" / "vosk"
+    if not vosk_model_path.exists():
+        raise SetupError(
+            "Vosk model not found. "
+            "Please download the model using `sprc download vosk` command."
+        )
 
     SAMPLE_RATE = 16_000
-    rec = KaldiRecognizer(Model("model"), SAMPLE_RATE)
+    rec = KaldiRecognizer(Model(str(vosk_model_path)), SAMPLE_RATE)
 
     rec.AcceptWaveform(
         audio_data.get_raw_data(convert_rate=SAMPLE_RATE, convert_width=2)