From aca449349a9d051ff956d74c518ab8137a999ab8 Mon Sep 17 00:00:00 2001
From: misyaguziya <misyaguziya@gmail.com>
Date: Fri, 30 Jun 2023 11:19:18 +0900
Subject: [PATCH] tempfile.mkstemp to io.BytesIO()

---
 AudioTranscriber.py | 22 +++++++++++--------
 test_main.py        | 53 ++++++++++++++++++++++++---------------------
 2 files changed, 41 insertions(+), 34 deletions(-)

diff --git a/AudioTranscriber.py b/AudioTranscriber.py
index 715e065b..dfdffd8a 100644
--- a/AudioTranscriber.py
+++ b/AudioTranscriber.py
@@ -1,5 +1,5 @@
+import io
 import os
-from io import BytesIO
 import tempfile
 import threading
 import wave
@@ -8,7 +8,7 @@ from datetime import timedelta
 import pyaudiowpatch as pyaudio
 
 PHRASE_TIMEOUT = 3.05
-MAX_PHRASES = 5
+MAX_PHRASES = 2
 
 class AudioTranscriber:
     def __init__(self, speaker, source, language):
@@ -34,14 +34,15 @@ class AudioTranscriber:
 
             text = ''
             try:
-                fd, path = tempfile.mkstemp(suffix=".wav")
-                os.close(fd)
-                audio_data = self.audio_sources["process_data_func"](path)
+                # fd, path = tempfile.mkstemp(suffix=".wav")
+                # os.close(fd)
+                audio_data = self.audio_sources["process_data_func"]()
                 text = self.audio_recognizer.recognize_google(audio_data, language=self.language)
             except Exception as e:
                 pass
             finally:
-                os.unlink(path)
+                pass
+                # os.unlink(path)
 
             if text != '':
                 self.update_transcript(text)
@@ -61,14 +62,16 @@ class AudioTranscriber:
         audio_data = sr.AudioData(self.audio_sources["last_sample"], self.audio_sources["sample_rate"], self.audio_sources["sample_width"])
         return audio_data
 
-    def process_speaker_data(self, path):
-        with wave.open(path, 'wb') as wf:
+    def process_speaker_data(self):
+        temp_file = io.BytesIO()
+        with wave.open(temp_file, 'wb') as wf:
             wf.setnchannels(self.audio_sources["channels"])
             p = pyaudio.PyAudio()
             wf.setsampwidth(p.get_sample_size(pyaudio.paInt16))
             wf.setframerate(self.audio_sources["sample_rate"])
             wf.writeframes(self.audio_sources["last_sample"])
-        with sr.AudioFile(path) as source:
+        temp_file.seek(0)
+        with sr.AudioFile(temp_file) as source:
             audio = self.audio_recognizer.record(source)
         return audio
 
@@ -84,6 +87,7 @@ class AudioTranscriber:
             transcript[0] = text
 
     def get_transcript(self):
+        print(self.transcript_data)
         if len(self.transcript_data) > 0:
             text = self.transcript_data.pop(-1)
         else:
diff --git a/test_main.py b/test_main.py
index 8821a081..b8f91074 100644
--- a/test_main.py
+++ b/test_main.py
@@ -5,33 +5,36 @@ import AudioTranscriber
 import AudioRecorder
 import audio_utils
 
-mic_audio_queue = queue.Queue()
-mic_device = audio_utils.get_default_input_device()
-mic_audio_recorder = AudioRecorder.SelectedMicRecorder(mic_device)
-mic_audio_recorder.record_into_queue(mic_audio_queue)
+try:
+    mic_audio_queue = queue.Queue()
+    mic_device = audio_utils.get_default_input_device()
+    mic_audio_recorder = AudioRecorder.SelectedMicRecorder(mic_device)
+    mic_audio_recorder.record_into_queue(mic_audio_queue)
 
-mic_transcriber = AudioTranscriber.AudioTranscriber(speaker=False, source=mic_audio_recorder.source, language="ja-JP")
-mic_transcribe = threading.Thread(target=mic_transcriber.transcribe_audio_queue, args=(mic_audio_queue,))
-mic_transcribe.daemon = True
-mic_transcribe.start()
+    mic_transcriber = AudioTranscriber.AudioTranscriber(speaker=False, source=mic_audio_recorder.source, language="ja-JP")
+    mic_transcribe = threading.Thread(target=mic_transcriber.transcribe_audio_queue, args=(mic_audio_queue,))
+    mic_transcribe.daemon = True
+    mic_transcribe.start()
 
-time.sleep(2)
+    time.sleep(2)
 
-spk_audio_queue = queue.Queue()
-spk_device = audio_utils.get_default_output_device()
-spk_audio_recorder = AudioRecorder.SelectedSpeakerRecorder(spk_device)
-spk_audio_recorder.record_into_queue(spk_audio_queue)
+    spk_audio_queue = queue.Queue()
+    spk_device = audio_utils.get_default_output_device()
+    spk_audio_recorder = AudioRecorder.SelectedSpeakerRecorder(spk_device)
+    spk_audio_recorder.record_into_queue(spk_audio_queue)
 
-spk_transcriber = AudioTranscriber.AudioTranscriber(speaker=True, source=spk_audio_recorder.source, language="ja-JP")
-spk_transcribe = threading.Thread(target=spk_transcriber.transcribe_audio_queue, args=(spk_audio_queue,))
-spk_transcribe.daemon = True
-spk_transcribe.start()
+    spk_transcriber = AudioTranscriber.AudioTranscriber(speaker=True, source=spk_audio_recorder.source, language="ja-JP")
+    spk_transcribe = threading.Thread(target=spk_transcriber.transcribe_audio_queue, args=(spk_audio_queue,))
+    spk_transcribe.daemon = True
+    spk_transcribe.start()
 
-while True:
-    text = mic_transcriber.get_transcript()
-    if len(text) > 0:
-        print("mic:", text)
-    text = spk_transcriber.get_transcript()
-    if len(text) > 0:
-        print("spk:", text)
-    time.sleep(0.1)
\ No newline at end of file
+    while True:
+        text = mic_transcriber.get_transcript()
+        if len(text) > 0:
+            print("mic:", text)
+        text = spk_transcriber.get_transcript()
+        if len(text) > 0:
+            print("spk:", text)
+        time.sleep(0.1)
+except Exception as e:
+    print(e)
\ No newline at end of file