1. add a new streamlit ui to preview manifest

2. implement rpcy transcription client for files

jasper/client.py

@@ -2,6 +2,10 @@ import os
 import logging
 import rpyc
 from functools import lru_cache
+import typer
+from pathlib import Path
+
+app = typer.Typer()
 
 logging.basicConfig(
     level=logging.INFO, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
@@ -19,3 +23,28 @@ def transcribe_gen(asr_host=ASR_HOST, asr_port=ASR_PORT):
     asr = rpyc.connect(asr_host, asr_port).root
     logger.info(f"connected to asr server successfully")
     return asr.transcribe
+
+
+@app.command()
+def transcribe_file(audio_file: Path):
+    from pydub import AudioSegment
+
+    transcriber = transcribe_gen()
+    aud_seg = (
+        AudioSegment.from_file_using_temporary_files(audio_file)
+        .set_channels(1)
+        .set_sample_width(2)
+        .set_frame_rate(24000)
+    )
+    tscript_file_path = audio_file.with_suffix(".txt")
+    transcription = transcriber(aud_seg.raw_data)
+    with open(tscript_file_path, "w") as tf:
+        tf.write(transcription)
+
+
+def main():
+    app()
+
+
+if __name__ == "__main__":
+    main()

jasper/data/utils.py

@@ -225,6 +225,12 @@ class ExtendedPath(type(Path())):
         with self.open("r") as jf:
             return json.load(jf)
 
+    def read_jsonl(self):
+        print(f"reading jsonl from {self}")
+        with self.open("r") as jf:
+            for l in jf.readlines():
+                yield json.loads(l)
+
     def write_json(self, data):
         print(f"writing json to {self}")
         self.parent.mkdir(parents=True, exist_ok=True)