pygpt-net 2.6.30__py3-none-any.whl → 2.6.31__py3-none-any.whl

pygpt_net/CHANGELOG.txt

@@ -1,3 +1,11 @@
+2.6.31 (2025-09-01)
+
+- Chat with Audio mode renamed to Realtime + audio.
+- Added support for real-time audio models from OpenAI (Realtime API) and Google (Live API), featuring real-time audio integration (beta).
+- Introduced new predefined models: gpt-realtime, gpt-4o-realtime-preview, and gemini-2.5-flash-preview-native-audio-dialog.
+- Included Google Gen AI audio input and output providers in the Audio Input/Output plugins.
+- Added URL Context remote tool support in Google Gen AI.
+
 2.6.30 (2025-08-29)
 
 - Added native Google GenAI API support (beta); live audio is not supported yet (#132).

pygpt_net/__init__.py

@@ -6,15 +6,15 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.08.29 00:00:00                  #
+# Updated Date: 2025.09.01 00:00:00                  #
 # ================================================== #
 
 __author__ = "Marcin Szczygliński"
 __copyright__ = "Copyright 2025, Marcin Szczygliński"
 __credits__ = ["Marcin Szczygliński"]
 __license__ = "MIT"
-__version__ = "2.6.30"
-__build__ = "2025-08-29"
+__version__ = "2.6.31"
+__build__ = "2025-09-01"
 __maintainer__ = "Marcin Szczygliński"
 __github__ = "https://github.com/szczyglis-dev/py-gpt"
 __report__ = "https://github.com/szczyglis-dev/py-gpt/issues"

pygpt_net/app.py

@@ -171,10 +171,12 @@ from pygpt_net.provider.audio_input.openai_whisper import OpenAIWhisper
 from pygpt_net.provider.audio_input.openai_whisper_local import OpenAIWhisperLocal
 from pygpt_net.provider.audio_input.google_speech_recognition import GoogleSpeechRecognition
 from pygpt_net.provider.audio_input.google_cloud_speech_recognition import GoogleCloudSpeechRecognition
+from pygpt_net.provider.audio_input.google_genai import GoogleGenAIAudioInput
 from pygpt_net.provider.audio_input.bing_speech_recognition import BingSpeechRecognition
 from pygpt_net.provider.audio_output.openai_tts import OpenAITextToSpeech
 from pygpt_net.provider.audio_output.ms_azure_tts import MSAzureTextToSpeech
 from pygpt_net.provider.audio_output.google_tts import GoogleTextToSpeech
+from pygpt_net.provider.audio_output.google_genai_tts import GoogleGenAITextToSpeech
 from pygpt_net.provider.audio_output.eleven_labs import ElevenLabsTextToSpeech
 
 # web search engine providers
@@ -318,10 +320,12 @@ def run(**kwargs):
     launcher.add_audio_input(OpenAIWhisperLocal())
     launcher.add_audio_input(GoogleSpeechRecognition())
     launcher.add_audio_input(GoogleCloudSpeechRecognition())
+    launcher.add_audio_input(GoogleGenAIAudioInput())
     launcher.add_audio_input(BingSpeechRecognition())
     launcher.add_audio_output(OpenAITextToSpeech())
     launcher.add_audio_output(MSAzureTextToSpeech())
     launcher.add_audio_output(GoogleTextToSpeech())
+    launcher.add_audio_output(GoogleGenAITextToSpeech())
     launcher.add_audio_output(ElevenLabsTextToSpeech())
 
     # register custom audio providers

pygpt_net/controller/__init__.py

@@ -1,4 +1,4 @@
-# !/usr/bin/env python3
+#!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 # ================================================== #
 # This file is a part of PYGPT package               #
@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.08.24 23:00:00                  #
+# Updated Date: 2025.08.30 06:00:00                  #
 # ================================================== #
 
 from pygpt_net.controller.access import Access
@@ -34,6 +34,7 @@ from pygpt_net.controller.model import Model
 from pygpt_net.controller.notepad import Notepad
 from pygpt_net.controller.painter import Painter
 from pygpt_net.controller.plugins import Plugins
+from pygpt_net.controller.realtime import Realtime
 from pygpt_net.controller.presets import Presets
 from pygpt_net.controller.settings import Settings
 from pygpt_net.controller.theme import Theme
@@ -76,6 +77,7 @@ class Controller:
         self.painter = Painter(window)
         self.plugins = Plugins(window)
         self.presets = Presets(window)
+        self.realtime = Realtime(window)
         self.settings = Settings(window)
         self.theme = Theme(window)
         self.tools = Tools(window)
@@ -108,6 +110,7 @@ class Controller:
         self.attachment.setup()
         self.camera.setup_ui()
         self.access.setup()
+        self.realtime.setup()
 
     def post_setup(self):
         """Post-setup, after plugins are loaded"""

pygpt_net/controller/audio/audio.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.08.27 07:00:00                  #
+# Updated Date: 2025.08.31 23:00:00                  #
 # ================================================== #
 
 import os
@@ -43,6 +43,30 @@ class Audio:
         if self.window.core.config.get("audio.input.continuous", False):
             self.window.ui.plugin_addon['audio.input.btn'].continuous.setChecked(True)
 
+        if self.window.core.config.get("audio.input.auto_turn", False):
+            self.window.ui.nodes['audio.auto_turn'].box.setChecked(True)
+
+    def execute_input_stop(self):
+        """Execute input stop (from UI)"""
+        self.window.dispatch(Event(Event.AUDIO_INPUT_RECORD_TOGGLE, {
+            "state": False,
+            "auto": True,  # do not emit manual event
+        }))
+
+    def is_recording(self) -> bool:
+        """
+        Check if audio input is recording
+
+        :return: True if recording
+        """
+        return self.window.core.plugins.get("audio_input").is_recording()
+
+    def toggle_auto_turn(self):
+        """Toggle auto turn setting"""
+        value = self.window.ui.nodes['audio.auto_turn'].box.isChecked()
+        self.window.core.config.set("audio.input.auto_turn", value)
+        self.window.core.config.save()
+
     def toggle_input(
             self,
             state: bool,

pygpt_net/controller/audio/ui.py

@@ -183,7 +183,7 @@ class UI:
         """
         self.recording = True
         self.window.ui.nodes['input'].set_icon_state("mic", True)
-        if mode == "input":
+        if mode in ["input", "realtime"]:
             self.window.controller.chat.common.lock_input()
             return
         btn = self.get_input_btn() if mode == 'input' else self.get_input_control_btn()
@@ -198,7 +198,7 @@ class UI:
         """
         self.recording = False
         self.window.ui.nodes['input'].set_icon_state("mic", False)
-        if mode == "input":
+        if mode in ["input", "realtime"]:
             self.window.controller.chat.common.unlock_input()
             return
         btn = self.get_input_btn() if mode == 'input' else self.get_input_control_btn()

pygpt_net/controller/chat/audio.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.12.14 18:00:00                  #
+# Updated Date: 2025.08.30 06:00:00                  #
 # ================================================== #
 
 import base64
@@ -40,19 +40,12 @@ class Audio:
         """Update input/output audio"""
         mode = self.window.core.config.get("mode")
         if mode == MODE_AUDIO:
-            if not self.window.controller.audio.is_output_enabled():
-                self.window.controller.audio.enable_output()
-                self.tmp_output = True
-            else:
-                self.tmp_output = False
             if not self.window.controller.audio.is_input_enabled():
                 self.window.controller.audio.enable_input()
                 self.tmp_input = True
             else:
                 self.tmp_input = False
         else:
-            if self.tmp_output:
-                self.window.controller.audio.disable_output()
             if self.tmp_input:
                 self.window.controller.audio.disable_input()
 

pygpt_net/controller/chat/common.py

@@ -10,12 +10,13 @@
 # ================================================== #
 
 import os
+from typing import Any
 
 from PySide6.QtGui import QTextCursor
 from PySide6.QtWidgets import QFileDialog, QApplication
 
 from pygpt_net.core.events import Event, AppEvent, RenderEvent, KernelEvent
-from pygpt_net.core.types import MODE_ASSISTANT
+from pygpt_net.core.types import MODE_ASSISTANT, MODE_AUDIO
 from pygpt_net.item.ctx import CtxItem
 from pygpt_net.item.model import ModelItem
 from pygpt_net.utils import trans
@@ -119,6 +120,17 @@ class Common:
         else:
             self.window.ui.config['global']['img_raw'].setChecked(False)
 
+        # image resolution
+        resolution = self.window.core.config.get('img_resolution', '1024x1024')
+        self.window.controller.config.apply_value(
+            parent_id="global",
+            key="img_resolution",
+            option=self.window.core.image.get_resolution_option(),
+            value=resolution,
+        )
+        if not self.initialized:
+            self.window.ui.add_hook("update.global.img_resolution", self.hook_update)
+
         # set focus to input
         self.window.ui.nodes['input'].setFocus()
         self.initialized = True
@@ -257,7 +269,7 @@ class Common:
             self.window.controller.access.voice.stop_recording(timeout=True)
 
         if self.window.core.plugins.get("audio_input").handler_simple.is_recording:
-            self.window.core.plugins.get("audio_input").handler_simple.stop_recording(timeout=False)
+            self.window.dispatch(Event(Event.AUDIO_INPUT_RECORD_TOGGLE))
             return
 
         # stop audio output if playing
@@ -275,7 +287,8 @@ class Common:
         """
         # don't unlock input and leave stop btn if assistant mode or if agent/autonomous is enabled
         # send btn will be unlocked in agent mode on stop
-        if self.can_unlock(ctx):
+        mode = self.window.core.config.get('mode')
+        if self.can_unlock(ctx) and mode != MODE_AUDIO:
             if not self.window.controller.kernel.stopped():
                 self.unlock_input()  # unlock input
                 return True
@@ -452,6 +465,19 @@ class Common:
         else:
             self.img_enable_raw()
 
+    def hook_update(self, key: str, value: Any, caller, *args, **kwargs):
+        """
+        Hook for updating image resolution
+
+        :param key: config key
+        :param value: new value
+        :param caller: caller object
+        """
+        if key == "img_resolution":
+            if not value:
+                return
+            self.window.core.config.set('img_resolution', value)
+
     def save_text(
             self,
             text: str,