audio compression

README.md

@@ -1,5 +1,5 @@
 ## Introduction
-This is a voice transcription bot for telegram. It runs Python 3 in a serverless AWS Lambda instance through [Chalice](https://github.com/aws/chalice) using [FFMpeg](https://ffmpeg.org/), [pyTelegramBotAPI](https://pypi.org/project/pyTelegramBotAPI/) and [Wit.AI](https://wit.ai/).
+This is a lightweight voice transcription bot for telegram. It runs Python 3 in a serverless AWS Lambda instance through [Chalice](https://github.com/aws/chalice) using [FFMpeg](https://ffmpeg.org/), [pyTelegramBotAPI](https://pypi.org/project/pyTelegramBotAPI/) and [Wit.AI](https://wit.ai/).
 
 ## Running
 ```sh

src/app.py

@@ -74,7 +74,7 @@ def webhook() -> None:
                         #&context=%7B%22locale%22%3A%22pt_BR%22%7D
                         'https://api.wit.ai/dictation',
                         headers = {
-                            'Content-Type': 'audio/mpeg',
+                            'Content-Type': file_utils.CONTENT_TYPE,
                             'Authorization': f'Bearer {wit_token}'
                         },
                         data = f,

src/chalicelib/file_utils.py

@@ -4,9 +4,10 @@
 MAX_DURATION = 300
 MAX_SIZE = 20 * 1024 * 1024 #MB
 
+CONTENT_TYPE = 'audio/mpeg'
+
 OGG_FILE_NAME = '/tmp/audio.ogg'
 MP3_FILE_NAME = '/tmp/audio.mp3'
-COMPRESSED_FILE_NAME = '/tmp/compressed.mp3'
 AAC_FILE_NAME = '/tmp/audio.aac'
 MP4_FILE_NAME = '/tmp/video.mp4'
 OUT_FILE_NAME = '/tmp/out%03d.mp3'
@@ -25,14 +26,15 @@ def __get_file_id(file) -> Union[str, None]:
 
 def __get_voice_file(voice) -> Union[str, None]:
     import subprocess
+    import os
 
     file_id = __get_file_id(voice)
     if file_id is None:
         return
 
     __download_file(file_id, OGG_FILE_NAME)
 
-    subprocess.Popen(['ffmpeg', '-y', '-i', OGG_FILE_NAME, MP3_FILE_NAME]).wait()
+    subprocess.Popen(['ffmpeg', '-y', '-i', OGG_FILE_NAME, '-map', '0:a:0', '-b:a', '41k', MP3_FILE_NAME]).wait()
 
     return MP3_FILE_NAME
 
@@ -57,18 +59,6 @@ def __split_file(file: str):
     subprocess.Popen(['ffmpeg', '-y', '-i', file, '-f', 'segment', '-segment_time', str(MAX_DURATION), '-c', 'copy', OUT_FILE_NAME]).wait()
     return glob.glob(OUT_BLOB)
 
-def __compress(file: str):
-    import os
-    import subprocess
-
-    stats = os.stat(file)
-    print(f'Original file size: {stats.st_size}')
-
-    subprocess.Popen(['ffmpeg', '-y', '-i', file, '-map', '0:a:0', '-b:a', '96k', COMPRESSED_FILE_NAME]).wait()
-
-    stats = os.stat(file)
-    print(f'Compressed file size: {stats.st_size}')
-
 def get_files(message) -> Union[list[str], None]:
     voice = message.get('voice')
     file = None
@@ -83,7 +73,6 @@ def get_files(message) -> Union[list[str], None]:
         duration = video.get('duration')
 
     if file:
-        #__compress(file)
         if duration and duration <= MAX_DURATION:
             return [file]
         else: