Events

voice_recognition.py

@@ -10,13 +10,47 @@ aai.settings.api_key = "aa9962f0088a449a9c4ab2361e96cc08"
 discord.opus._load_default()
 
 # URL of the file to transcribe
-FILE_URL = "https://github.com/AssemblyAI-Community/audio-examples/raw/main/20230607_me_canadian_wildfires.mp3"
-
 # You can also transcribe a local file by passing in a file path
-# FILE_URL = './path/to/file.mp3'
+# FILE_URL = './path/to/file.mWp3
 
 logger = logging.getLogger("discord")
+location = "/home/pi/Conjurer/"
+class Events(commands.Cog):
+    def __init__(self, bot):
+        self.bot = bot
 
+    @commands.Cog.listener()
+    async def on_voice_state_update(self, before, after):
+        logger.info("before %s", before)
+        logger.info("after %s", after)
+
+
+class UserTranscript():
+    def __init__(self, user_id):
+
+        self.username = str(user_id)
+        self.present_file_id = 0
+        self.file_name_past = None
+
+        self.file_name_present = location + self.username + str(self.present_file_id)
+        self.transcript_file_present = wave.Wave_write = wave.open(self.file_name_present, 'wb')
+
+        self.file_name_future = location + self.username + str(self.present_file_id+1)
+        self.transcript_file_future = wave.Wave_write = wave.open(self.file_name_future, 'wb')
+    def rotate(self):
+        self.transcript_file_present.close()
+        self.name_file_past = self.file_name_present
+        self.present_file_id += 1
+        self.file_name_present = self.file_name_future
+        self.transcript_file_present = self.transcript_file_future
+        self.transcript_file_future = location + self.username + str(self.present_file_id+1)
+        self.transcript_file_future = wave.Wave_write = wave.open(self.file_name_future, 'wb')
+        #send past to transcription
+
+    def cleanup(self):
+        self.transcript_file_present.close()
+        self.transcript_file_future.close()
+        #send present to transcription
 
 class SRBuffer(voice_recv.AudioSink):
     """Endpoint AudioSink that generates a wav file.
@@ -27,22 +61,10 @@ class SRBuffer(voice_recv.AudioSink):
     SAMPLE_WIDTH = OpusDecoder.SAMPLE_SIZE // OpusDecoder.CHANNELS
     SAMPLING_RATE = OpusDecoder.SAMPLING_RATE
 
-    ##on member join - 5 plików buforowych na człowieka otwarty jest tylko aktualny i następny
+    #on member join dodajemytypa do listy
     #on member disconnect - dropujemy go
 
 
-    #        event: BasicSinkWriteCB,
-    #        rtcp_event: Optional[BasicSinkWriteRTCPCB] = None,
-    #        self.cb = event
-    #        self.cb_rtcp = rtcp_event
-    #   def write(self, user: Optional[User], data: VoiceData) -> None:
-    #       self.cb(user, data)
-    #   @AudioSink.listener()
-    #   def on_rtcp_packet(self, packet: RTCPPacket, guild: discord.Guild) -> None:
-    #       self.cb_rtcp(packet) if self.cb_rtcp else None
-
-
-
     def __init__(self, destination):
         super().__init__()
 
@@ -50,7 +72,7 @@ class SRBuffer(voice_recv.AudioSink):
         self._file.setnchannels(self.CHANNELS)
         self._file.setsampwidth(self.SAMPLE_WIDTH)
         self._file.setframerate(self.SAMPLING_RATE)
-        self.user_list = []
+        self.user_list = {}
 
     def wants_opus(self) -> bool:
         return False
@@ -67,22 +89,18 @@ class SRBuffer(voice_recv.AudioSink):
             logger.warning("WaveSink got error closing file on cleanup", exc_info=True)
 
 
-
 class Transcriber(commands.Cog):
     def __init__(self, bot):
-        def callback(user, data):
-            logger.info(f"Got packet from {user}")
-
         self.bot = bot
         self._last_member = None
         self.wsink = SRBuffer(destination="/home/pi/Conjurer/wav.wav")
 
-    async def transcribe():
-        config = aai.TranscriptionConfig(speaker_labels=True)
+    async def transcribe(file_url):
+        config = aai.TranscriptionConfig(speaker_labels=True, language_code="pl")
 
         transcriber = aai.Transcriber()
         transcript = transcriber.transcribe(
-        FILE_URL,
+        file_url,
         config=config
         )
         for utterance in transcript.utterances:
@@ -101,7 +119,7 @@ class Transcriber(commands.Cog):
 
 async def setup(bot):
     await bot.add_cog(Transcriber(bot))
-
+    await bot.add_cog(Events(bot))
 
 #1. zapisuj kwestie człowieka do pliku w którym będzie można stwierdzić kto co powiedział (callback z basicaudio + write z wavesinka). Zamykaj plik i wysyłaj do transkrypcji w momencie ciszy dłuższej niż 0.5s. Jeśli człowiek nadaje cały czas dawaj sygnał że nie można go transkrybować - i wyłaczaj zapis. 
 #2. Transkrypcja to abstrakt - w zależności od tego która metoda jest włączona wysyła do odpowiedniego silnika, silniki offline odpalam na activcomie. Zaczynamy od assemblyai bo najlepiej wspiera polski. Potem zobaczymy co dalej.