feat: fixed direct recording to webservice

main.py

@@ -360,14 +360,16 @@ def main(page):
 
         set_transcribe_ready(False)
 
-    def paralyze_ui():
+    def paralyze_ui(spinner: bool = True, disable_recording_button: bool = True):
         model_size_select.current.disabled = True
         model_device_select.current.disabled = True
         # model_bits_select.current.disabled = True
         model_load_unload_button.current.disabled = True
-        processing_spinner.current.visible = True
+        processing_spinner.current.visible = spinner
         current_mode_select.current.disabled = True
 
+        record_button.current.disabled = disable_recording_button
+
         model_load_unload_button.current.icon = ft.icons.CLOSE
         model_load_unload_button.current.disabled = False
         for btn in transcribe_buttons:
@@ -392,6 +394,8 @@ def main(page):
             model_load_unload_button.current.icon = ft.icons.CLOSE
             model_load_unload_button.current.disabled = False
 
+            record_button.current.disabled = False
+
             if mm.is_model_loaded():
                 current_mode_select.current.disabled = True
         else:
@@ -406,6 +410,8 @@ def main(page):
             processing_spinner.current.visible = False
             current_mode_select.current.disabled = False
 
+            record_button.current.disabled = True
+
         page.update()
 
     def on_url_input(e):
@@ -442,15 +448,23 @@ def main(page):
 
             recorded_audio = b"".join(sound_chunks)
 
+            set_transcribe_ready(False)
+
             transcribe(recorded_audio)
 
             recording = False
 
             # sound = pygame.mixer.Sound(buffer=recorded_audio)  # doesn't work because sampling rate is wrong
+            record_button.current.bgcolor = "0x000000FF"
 
-            print("playing back recorded sound")
+            set_transcribe_ready(True)
+
+            print("done")
             # sound.play()
         else:
+            if not transcribe_ready:
+                print("Can't record, not ready")
+                return
             print("Starting Recording...")
             recording = True
 
@@ -472,6 +486,9 @@ def main(page):
 
             rec_stream.start_stream()
 
+            record_button.current.bgcolor = "0xFFFF4444"
+            paralyze_ui(spinner=False, disable_recording_button=False)
+
     def find_recordingdevice_tuple_by_name(search_name: str) -> typing.Tuple[int, str] | None:
         return next(((device_id, name) for device_id, name in capture_devices if name == search_name))
 
@@ -594,5 +611,8 @@ def main(page):
         ], expand=True),
     )
 
+    # refresh all values, and make sure the right stuff is shown
+    mode_select()
+
 
 ft.app(target=main)

pyproject.toml

@@ -9,15 +9,13 @@ readme = "README.md"
 [tool.poetry.dependencies]
 python = ">=3.11, <3.13"
 flet = "^0.10.3"
-faster-whisper = "^0.9.0"
+faster-whisper = "0.10.0"
 pygame = "^2.5.2"
-torch = "2.0.0"
 requests = "^2.31.0"
 validators = "^0.22.0"
 pyinstaller = "^6.1.0"
-pysdl2 = "^0.9.16"
-pysdl2-dll = "^2.28.4"
 pyaudio = "^0.2.13"
+pydub = "^0.25.1"
 
 
 [build-system]

utils.py

@@ -3,6 +3,9 @@ import os
 
 from typing import DefaultDict, Dict, List
 
+from pydub import AudioSegment
+import io
+
 
 def tree() -> DefaultDict:
     return defaultdict(tree)
@@ -44,3 +47,15 @@ def defaultdict_to_dict(d: defaultdict) -> dict:
     if isinstance(d, defaultdict):
         d = {k: defaultdict_to_dict(v) for k, v in d.items()}
     return d
+
+
+def convert_to_mp3(audio_data: bytes, sample_width: int, frame_rate: int, channels: int) -> bytes:
+    audio = AudioSegment.from_raw(
+        io.BytesIO(audio_data),
+        sample_width=sample_width,
+        frame_rate=frame_rate,
+        channels=channels
+    )
+    mp3_buffer = io.BytesIO()
+    audio.export(mp3_buffer, format="mp3")
+    return mp3_buffer.getvalue()

whisper_webservice_interface.py

@@ -2,6 +2,8 @@ from typing import Optional, Union, Dict, Any
 
 import requests
 
+from utils import convert_to_mp3
+
 
 def send_asr_request(url: str, audio_file_path_or_bytes: str | bytes, task: Optional[str] = None, language: Optional[str] = None,
                      initial_prompt: Optional[str] = None, encode: Optional[bool] = None,
@@ -28,7 +30,7 @@ def send_asr_request(url: str, audio_file_path_or_bytes: str | bytes, task: Opti
         with open(audio_file_path_or_bytes, 'rb') as f:
             audio_file = f.read()
     else:
-        audio_file = audio_file_path_or_bytes
+        audio_file = convert_to_mp3(audio_file_path_or_bytes, sample_width=2, frame_rate=16000, channels=1)
 
     files = {
         'audio_file': audio_file