pygpt-net 2.4.49__py3-none-any.whl → 2.4.51__py3-none-any.whl

pygpt_net/core/audio/capture.py

@@ -0,0 +1,349 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+# ================================================== #
+# This file is a part of PYGPT package               #
+# Website: https://pygpt.net                         #
+# GitHub:  https://github.com/szczyglis-dev/py-gpt   #
+# MIT License                                        #
+# Created By  : Marcin Szczygliński                  #
+# Updated Date: 2025.01.17 02:00:00                  #
+# ================================================== #
+
+import time
+
+import numpy as np
+import wave
+
+from PySide6.QtMultimedia import QAudioFormat, QMediaDevices, QAudioSource
+from PySide6.QtCore import QTimer
+
+class Capture:
+    def __init__(self, window=None):
+        """
+        Audio input capture core
+
+        :param window: Window instance
+        """
+        self.window = window
+        self.audio_source = None
+        self.audio_io_device = None
+        self.frames = []
+        self.actual_audio_format = None
+        self.path = None
+        self.disconnected = False
+        self.loop = False
+        self.stop_callback = None
+        self.start_time = 0
+        self.devices = []
+        self.selected_device = None
+        self.bar = None
+        self.check_audio_devices()
+
+    def set_path(self, path: str):
+        """
+        Set audio input file path
+
+        :param path: file path
+        """
+        self.path = path
+
+    def set_bar(self, bar):
+        """
+        Set audio level bar
+
+        :param bar: audio level bar
+        """
+        self.bar = bar
+
+    def check_audio_devices(self):
+        """Check audio input devices"""
+        self.devices = QMediaDevices.audioInputs()
+        if not self.devices:
+            # no devices found
+            self.selected_device = None
+            print("No audio input devices found.")
+        else:
+            # set the first device as default
+            self.selected_device = self.devices[0]
+
+    def device_changed(self, index: int):
+        """
+        Change audio input device
+
+        :param index: device index
+        """
+        if 0 <= index < len(self.devices):
+            self.selected_device = self.devices[index]
+        else:
+            self.selected_device = None
+
+    def prepare_device(self):
+        """Set the current audio input device"""
+        device_id = int(self.window.core.config.get('audio.input.device', 0))
+        self.device_changed(device_id)
+
+    def start(self):
+        """
+        Start audio input recording
+
+        :return: True if started
+        """
+        # Clear previous frames
+        self.frames = []
+
+        # Prepare selected device
+        self.prepare_device()
+        if not self.selected_device:
+            print("No audio input device selected")
+            return
+        if self.disconnected:
+            print("Audio source disconnected, please connect the audio source")
+            return False
+
+        # Prevent multiple recordings
+        if self.audio_source is not None:
+            return False
+
+        # Set up audio input and start recording
+        self.setup_audio_input()
+        self.start_time = time.time()
+        return True
+
+    def stop(self):
+        """
+        Stop audio input recording
+
+        :return: True if stopped
+        """
+        result = False
+        if self.audio_source is not None:
+            # Disconnect the readyRead signal
+            self.audio_io_device.readyRead.disconnect(self.process_audio_input)
+            self.audio_source.stop()
+            self.audio_source = None
+            self.audio_io_device = None
+
+            # Save frames to file (if any)
+            if self.frames:
+                self.save_audio_file(self.path)
+                result = True
+            else:
+                print("No audio data recorded")
+        return result
+
+    def has_source(self) -> bool:
+        """
+        Check if audio source is available
+
+        :return: True if available
+        """
+        if self.audio_source is not None:
+            return True
+        return False
+
+    def has_frames(self) -> bool:
+        """
+        Check if audio frames are available
+
+        :return: True if available
+        """
+        if self.frames:
+            return True
+        return False
+
+    def get_frames(self) -> list:
+        """
+        Get recorded audio frames
+
+        :return: list of frames
+        """
+        return self.frames
+
+    def setup_audio_input(self):
+        """Set up audio input device and start recording"""
+        if not self.selected_device:
+            print("No audio input device selected")
+            return
+
+        # Define audio format
+        audio_format = QAudioFormat()
+        audio_format.setSampleRate(int(self.window.core.config.get('audio.input.rate', 44100)))
+        audio_format.setChannelCount(int(self.window.core.config.get('audio.input.channels', 1)))
+        audio_format.setSampleFormat(QAudioFormat.SampleFormat.Int16)
+
+        # Select default audio input device
+        audio_input_device = self.selected_device
+
+        # Check if the format is supported
+        if not audio_input_device.isFormatSupported(audio_format):
+            print("Requested format not supported, using nearest format.")
+            audio_format = audio_input_device.preferredFormat()
+
+        # Store the actual format being used
+        self.actual_audio_format = audio_format
+
+        # Create QAudioSource object with the device and format
+        try:
+            self.audio_source = QAudioSource(audio_input_device, audio_format)
+        except Exception as e:
+            self.disconnected = True
+            print(f"Failed to create audio source: {e}")
+
+        # Start audio input and obtain the QIODevice
+        try:
+            self.audio_io_device = self.audio_source.start()
+            if self.audio_io_device is None:
+                raise Exception("Unable to access audio input device")
+        except Exception as e:
+            print(f"Failed to start audio input: {e}")
+            self.disconnected = True
+            self.audio_source = None
+            self.audio_io_device = None
+            return
+
+        # Connect the readyRead signal to process incoming data
+        self.audio_io_device.readyRead.connect(self.process_audio_input)
+
+    def process_audio_input(self):
+        """Process incoming audio data"""
+        # add seconds to stop timer
+        data = self.audio_io_device.readAll()
+        if data.isEmpty():
+            return
+
+        # Convert QByteArray to bytes
+        data_bytes = data.data()
+
+        # Append raw data to frames list for saving
+        self.frames.append(data_bytes)
+
+        # Determine the correct dtype and normalization factor
+        sample_format = self.actual_audio_format.sampleFormat()
+        dtype = self.get_dtype_from_sample_format(sample_format)
+        normalization_factor = self.get_normalization_factor(sample_format)
+
+        # Convert bytes to NumPy array of the appropriate type
+        samples = np.frombuffer(data_bytes, dtype=dtype)
+        if samples.size == 0:
+            return
+
+        # For unsigned formats, center the data
+        if sample_format == QAudioFormat.SampleFormat.UInt8:
+            samples = samples.astype(np.int16)
+            samples -= 128
+
+        # Compute RMS of the audio samples as float64 for precision
+        rms = np.sqrt(np.mean(samples.astype(np.float64) ** 2))
+
+        # Normalize RMS value based on the sample format
+        level = rms / normalization_factor
+
+        # Ensure level is within 0.0 to 1.0
+        level = min(max(level, 0.0), 1.0)
+
+        # Scale to 0-100
+        level_percent = level * 100
+
+        # Update the level bar widget
+        self.update_audio_level(level_percent)
+
+        # Handle loop recording
+        if self.loop and self.stop_callback is not None:
+            stop_interval = int(self.window.core.config.get('audio.input.stop_interval', 10))
+            current_time = time.time()
+            time_elapsed = current_time - self.start_time
+            if time_elapsed >= stop_interval:
+                # stop recording here, save audio chunk to WAV file, run transcription, and start recording again
+                self.start_time = current_time
+                QTimer.singleShot(0, self.stop_callback)  # required QTimer to prevent crash!!!
+
+    def update_audio_level(self, level: int):
+        """
+        Update the audio level bar
+
+        :param level: audio level
+        """
+        if self.bar is not None:
+            self.bar.setLevel(level)
+
+    def reset_audio_level(self):
+        """Reset the audio level bar"""
+        if self.bar is not None:
+            self.bar.setLevel(0)
+
+    def save_audio_file(self, filename: str):
+        """
+        Save the recorded audio frames to a WAV file
+
+        :param filename: output file name
+        """
+        # Define the parameters for the WAV file
+        channels = self.actual_audio_format.channelCount()
+        sample_size = self.actual_audio_format.bytesPerSample()
+        frame_rate = self.actual_audio_format.sampleRate()
+
+        # Open the WAV file
+        wf = wave.open(filename, 'wb')
+        wf.setnchannels(channels)
+        wf.setsampwidth(sample_size)
+        wf.setframerate(frame_rate)
+
+        # Write frames to the file
+        wf.writeframes(b''.join(self.frames))
+        wf.close()
+
+    def get_dtype_from_sample_format(self, sample_format):
+        """
+        Get the NumPy dtype corresponding to the QAudioFormat sample format
+
+        :param sample_format: QAudioFormat.SampleFormat
+        """
+        if sample_format == QAudioFormat.SampleFormat.UInt8:
+            return np.uint8
+        elif sample_format == QAudioFormat.SampleFormat.Int16:
+            return np.int16
+        elif sample_format == QAudioFormat.SampleFormat.Int32:
+            return np.int32
+        elif sample_format == QAudioFormat.SampleFormat.Float:
+            return np.float32
+        else:
+            raise ValueError("Unsupported sample format")
+
+    def get_normalization_factor(self, sample_format):
+        """
+        Get the normalization factor for the QAudioFormat sample format
+
+        :param sample_format: QAudioFormat.SampleFormat
+        """
+        if sample_format == QAudioFormat.SampleFormat.UInt8:
+            return 255.0
+        elif sample_format == QAudioFormat.SampleFormat.Int16:
+            return 32768.0
+        elif sample_format == QAudioFormat.SampleFormat.Int32:
+            return float(2 ** 31)
+        elif sample_format == QAudioFormat.SampleFormat.Float:
+            return 1.0
+        else:
+            raise ValueError("Unsupported sample format")
+
+    def check_audio_input(self) -> bool:
+        """
+        Check if default audio input device is working
+
+        :return: True if working
+        """
+        import pyaudio
+        try:
+            rate = 44100
+            channels = 1
+            p = pyaudio.PyAudio()
+            stream = p.open(format=pyaudio.paInt16,
+                            channels=channels,
+                            rate=rate,
+                            input=True)
+            stream.stop_stream()
+            stream.close()
+            p.terminate()
+            return True
+        except Exception as e:
+            return False

pygpt_net/data/config/config.json

@@ -1,8 +1,8 @@
 {
   "__meta__": {
-    "version": "2.4.49",
-    "app.version": "2.4.49",
-    "updated_at": "2025-01-16T00:00:00"
+    "version": "2.4.51",
+    "app.version": "2.4.51",
+    "updated_at": "2025-01-17T00:00:00"
   },
   "access.audio.event.speech": false,
   "access.audio.event.speech.disabled": [],
@@ -74,8 +74,10 @@
   "attachments_send_clear": true,
   "attachments_capture_clear": true,
   "audio.input.channels": 1,
+  "audio.input.continuous": false,
   "audio.input.device": "0",
   "audio.input.rate": 44100,
+  "audio.input.stop_interval": 10,
   "audio.transcribe.convert_video": true,
   "context_threshold": 200,
   "cmd": false,

pygpt_net/data/config/models.json

@@ -1,8 +1,8 @@
 {
     "__meta__": {
-        "version": "2.4.49",
-        "app.version": "2.4.49",
-        "updated_at": "2025-01-16T00:00:00"
+        "version": "2.4.51",
+        "app.version": "2.4.51",
+        "updated_at": "2025-01-17T00:00:00"
     },
     "items": {
         "claude-3-5-sonnet-20240620": {

pygpt_net/data/config/modes.json

@@ -1,8 +1,8 @@
 {
     "__meta__": {
-        "version": "2.4.49",
-        "app.version": "2.4.49",
-        "updated_at": "2025-01-16T00:00:00"
+        "version": "2.4.51",
+        "app.version": "2.4.51",
+        "updated_at": "2025-01-17T00:00:00"
     },
     "items": {
         "chat": {

pygpt_net/data/config/settings.json

@@ -1123,6 +1123,19 @@
         "step": 1,
         "advanced": false
     },
+    "audio.input.stop_interval": {
+        "section": "audio",
+        "type": "int",
+        "slider": true,
+        "label": "settings.audio.input.stop_interval",
+        "description": "settings.audio.input.stop_interval.desc",
+        "value": 10,
+        "min": 5,
+        "max": 120,
+        "multiplier": 1,
+        "step": 1,
+        "advanced": false
+    },
     "llama.idx.list": {
         "section": "llama-index",
         "type": "dict",

pygpt_net/data/locale/locale.de.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = Zauberwort erkannt!
 audio.magic_word.invalid = Kein Zauberwort :(
 audio.magic_word.please = Bitte das Zauberwort...
 audio.speak.btn = Mikrofon
+audio.speak.btn.continuous = Kontinuierliche Aufzeichnung
 audio.speak.btn.stop = Stopp
 audio.speak.btn.stop.tooltip = Klicken, um das Mikrofonhören zu beenden
 audio.speak.btn.tooltip = Klicken, um das Mikrofonhören zu starten
@@ -774,6 +775,8 @@ settings.audio.input.device = Audioeingabegerät
 settings.audio.input.device.desc = Wählen Sie das Audiogerät für die Mikrofoneingabe.
 settings.audio.input.rate = Abtastrate
 settings.audio.input.rate.desc = Abtastrate, Standard: 44100
+settings.audio.input.stop_interval = Intervall für kontinuierliche automatische Transkription
+settings.audio.input.stop_interval.desc = Intervall in Sekunden für automatisches Transkribieren eines Audioabschnitts, Standard: 10
 settings.check_updates = Beim Start nach Updates suchen
 settings.check_updates.bg = Im Hintergrund nach Updates suchen
 settings.cmd.field.desc = Aktivieren Sie das `{cmd}`-Werkzeug.

pygpt_net/data/locale/locale.en.ini

@@ -147,6 +147,7 @@ audio.magic_word.detected = Magic word detected!
 audio.magic_word.invalid = Not a magic word :(
 audio.magic_word.please = Magic word please...
 audio.speak.btn = Microphone
+audio.speak.btn.continuous = Continuous recording
 audio.speak.btn.stop = Stop
 audio.speak.btn.stop.tooltip = Click to stop microphone listening
 audio.speak.btn.tooltip = Click to start microphone listening
@@ -945,6 +946,8 @@ settings.audio.input.device = Audio Input Device
 settings.audio.input.device.desc = Select the audio device for Microphone input.
 settings.audio.input.rate = Sampling Rate
 settings.audio.input.rate.desc = Sampling rate, default: 44100
+settings.audio.input.stop_interval = Continuous recording auto-transcribe interval
+settings.audio.input.stop_interval.desc = Interval in seconds for auto-transcribe audio chunk, default: 10
 settings.check_updates = Check for Updates on start
 settings.check_updates.bg = Check for Updates in the background
 settings.cmd.config.collapse = JSON params

pygpt_net/data/locale/locale.es.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = ¡Palabra mágica detectada!
 audio.magic_word.invalid = No es una palabra mágica :(
 audio.magic_word.please = Por favor, una palabra mágica...
 audio.speak.btn = Micrófono
+audio.speak.btn.continuous = Grabación continua
 audio.speak.btn.stop = Detener
 audio.speak.btn.stop.tooltip = Haz clic para dejar de escuchar a través del micrófono
 audio.speak.btn.tooltip = Haz clic para empezar a escuchar a través del micrófono
@@ -773,6 +774,8 @@ settings.audio.input.device = Dispositivo de Entrada de Audio
 settings.audio.input.device.desc = Selecciona el dispositivo de audio para la entrada del micrófono.
 settings.audio.input.rate = Taux d'échantillonnage
 settings.audio.input.rate.desc = Taux d'échantillonnage, par défaut : 44100
+settings.audio.input.stop_interval = Intervalo de transcripción automática continua
+settings.audio.input.stop_interval.desc = Intervalo en segundos para transcribir automáticamente el fragmento de audio, predeterminado: 10
 settings.check_updates = Buscar actualizaciones al iniciar
 settings.check_updates.bg = Buscar actualizaciones en segundo plano
 settings.cmd.field.desc = Habilitar la herramienta `{cmd}`.

pygpt_net/data/locale/locale.fr.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = Mot magique détecté !
 audio.magic_word.invalid = Pas un mot magique :(
 audio.magic_word.please = S'il vous plaît, le mot magique...
 audio.speak.btn = Microphone
+audio.speak.btn.continuous = Enregistrement continu
 audio.speak.btn.stop = Arrêt
 audio.speak.btn.stop.tooltip = Cliquez pour arrêter l'écoute par le microphone
 audio.speak.btn.tooltip = Cliquez pour commencer l'écoute par le microphone
@@ -773,6 +774,8 @@ settings.audio.input.device = Périphérique d'Entrée Audio
 settings.audio.input.device.desc = Sélectionnez le périphérique audio pour l'entrée du microphone.
 settings.audio.input.rate = Taux d'échantillonnage
 settings.audio.input.rate.desc = Taux d'échantillonnage, par défaut : 44100
+settings.audio.input.stop_interval = Intervalle d'auto-transcription enregistrement continu
+settings.audio.input.stop_interval.desc = Intervalle en secondes pour la transcription automatique d'un morceau audio, par défaut : 10
 settings.check_updates = Vérifier les mises à jour au démarrage
 settings.check_updates.bg = Vérifier les mises à jour en arrière-plan
 settings.cmd.field.desc = Activer l'outil `{cmd}`.

pygpt_net/data/locale/locale.it.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = Parola magica rilevata!
 audio.magic_word.invalid = Non è una parola magica :(
 audio.magic_word.please = Per favore, parola magica...
 audio.speak.btn = Microfono
+audio.speak.btn.continuous = Registrazione continua
 audio.speak.btn.stop = Stop
 audio.speak.btn.stop.tooltip = Clicca per terminare l'ascolto tramite microfono
 audio.speak.btn.tooltip = Clicca per iniziare l'ascolto tramite microfono
@@ -774,6 +775,8 @@ settings.audio.input.device = Dispositivo di Ingresso Audio
 settings.audio.input.device.desc = Seleziona il dispositivo audio per l'ingresso del microfono.
 settings.audio.input.rate = Frequenza di campionamento
 settings.audio.input.rate.desc = Frequenza di campionamento, predefinito: 44100
+settings.audio.input.stop_interval = Intervallo di trascrizione automatica continua
+settings.audio.input.stop_interval.desc = Intervallo in secondi per la trascrizione automatica del segmento audio, predefinito: 10
 settings.check_updates = Controlla aggiornamenti all'avvio
 settings.check_updates.bg = Controlla gli aggiornamenti in background
 settings.cmd.field.desc = Abilita lo strumento `{cmd}`.

pygpt_net/data/locale/locale.pl.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = Wykryto magiczne słowo!
 audio.magic_word.invalid = To nie magiczne słowo :(
 audio.magic_word.please = Podaj magiczne słowo...
 audio.speak.btn = Mikrofon
+audio.speak.btn.continuous = Ciągłe nagrywanie
 audio.speak.btn.stop = Stop
 audio.speak.btn.stop.tooltip = Kliknij, aby zakończyć nasłuchiwanie przez mikrofon
 audio.speak.btn.tooltip = Kliknij, aby rozpocząć nasłuchiwanie przez mikrofon
@@ -774,6 +775,8 @@ settings.audio.input.device = Urządzenie Wejściowe Audio
 settings.audio.input.device.desc = Wybierz urządzenie audio dla wejścia mikrofonu.
 settings.audio.input.rate = Częstotliwość próbkowania
 settings.audio.input.rate.desc = Częstotliwość próbkowania, domyślnie: 44100
+settings.audio.input.stop_interval = Interwał automatycznego transkrybowania nagrania ciągłego
+settings.audio.input.stop_interval.desc = Interwał w sekundach dla automatycznego transkrybowania fragmentu audio, domyślnie: 10
 settings.check_updates = Sprawdź aktualizacje przy starcie
 settings.check_updates.bg = Sprawdź aktualizacje w tle
 settings.cmd.field.desc = Włącz narzędzie `{cmd}`.

pygpt_net/data/locale/locale.uk.ini

@@ -135,6 +135,7 @@ audio.magic_word.detected = Магічне слово виявлено!
 audio.magic_word.invalid = Не магічне слово :(
 audio.magic_word.please = Магічне слово, будь ласка...
 audio.speak.btn = Мікрофон
+audio.speak.btn.continuous = Безперервний запис
 audio.speak.btn.stop = Стоп
 audio.speak.btn.stop.tooltip = Натисніть, щоб закінчити слухання через мікрофон
 audio.speak.btn.tooltip = Натисніть, щоб почати слухання через мікрофон
@@ -773,6 +774,8 @@ settings.audio.input.device = Аудіо Вхідний Пристрій
 settings.audio.input.device.desc = Виберіть аудіо пристрій для введення мікрофона.
 settings.audio.input.rate = Частота дискретизації
 settings.audio.input.rate.desc = Частота дискретизації, за замовчуванням: 44100
+settings.audio.input.stop_interval = Інтервал автоматичного транскрибування безперервного запису
+settings.audio.input.stop_interval.desc = Інтервал у секундах для автоматичного транскрибування фрагмента аудіо, за замовчуванням: 10
 settings.check_updates = Перевіряти оновлення при запуску
 settings.check_updates.bg = Перевіряти оновлення у фоновому режимі
 settings.cmd.field.desc = Увімкнути інструмент `{cmd}`.

pygpt_net/data/locale/locale.zh.ini

@@ -136,6 +136,7 @@ audio.control.btn = 语音控制
 audio.magic_word.detected = 檢測到魔法詞！
 audio.magic_word.invalid= 不是魔法詞 :(
 audio.magic_word.please= 請說出魔法詞...
+audio.speak.btn.continuous = 连续录音
 audio.speak.btn.stop = 停止
 audio.speak.btn.stop.tooltip = 點擊停止麥克風聆聽
 audio.speak.btn.tooltip = 點擊開始麥克風聆聽
@@ -875,6 +876,8 @@ settings.audio.input.device = 音频输入设备
 settings.audio.input.device.desc = 选择用于麦克风输入的音频设备。
 settings.audio.input.rate = 采样率
 settings.audio.input.rate.desc = 采样率，默认: 44100
+settings.audio.input.stop_interval = 连续录音自动转录间隔
+settings.audio.input.stop_interval.desc = 自动转录音频片段的间隔（以秒为单位），默认：10
 settings.check_updates = 啟動時檢查更新
 settings.check_updates.bg = 在後台檢查更新
 settings.cmd.field.desc = 启用 `{cmd}` 工具。