feat: voice mode

[phact]

This pull request introduces a new voice assistant feature to the application, including backend API support and frontend integration. The most important changes include adding the voice_mode_router, creating a new WebSocket endpoint for voice interactions, and implementing the frontend components for voice recording and playback.

It also handles the case where the OpenAI key is missing and prompts the user.

Backend Changes:

• src/backend/base/langflow/api/router.py: Added voice_mode_router to the main API router. [1] [2]
• src/backend/base/langflow/api/v1/__init__.py: Imported and registered voice_mode_router. [1] [2]
• src/backend/base/langflow/api/v1/voice_mode.py: Created a new module for handling voice mode interactions, including WebSocket endpoints and functions for managing voice sessions.

Frontend Changes:

• src/frontend/src/components/VoiceAssistant/index.tsx: Implemented the VoiceAssistant component, which handles audio recording, WebSocket communication, and playback.
• src/frontend/src/components/VoiceAssistant/streamProcessor.ts: Added an AudioWorkletProcessor for processing audio streams.
• src/frontend/src/components/VoiceAssistant/utils.ts: Added utility functions for converting base64 audio data to Float32Array.
• src/frontend/src/modals/IOModal/components/chatView/chatInput/newChatInput.tsx: Integrated the VoiceAssistant component into the chat input view. [1] [2]
• src/frontend/src/modals/IOModal/index.tsx: Added the VoiceAssistant component to the I/O modal. [1] [2]

[codspeed-hq]

CodSpeed Performance Report

Merging #4642 will degrade performances by 38.94%

_{Comparing phact:voice_mode (55aad4d) with main (517961d)}

Summary

⚡ 1 improvements
❌ 2 regressions
✅ 12 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	phact:voice_mode	Change
❌	test_get_and_cache_all_types_dict	1.2 ms	2 ms	-38.94%
❌	test_successful_run_with_input_type_any	175.2 ms	261.1 ms	-32.92%
⚡	test_successful_run_with_input_type_text	333.3 ms	249 ms	+33.83%