Features

Drag & drop audio or video files and convert them to accurate text with one click. Handle meeting recordings, interviews, lectures, and more.

• Auto-selects the best of 4 backends: CUDA / OpenVINO / Vulkan / CPU
• Batch parallel processing for multiple files simultaneously
• Loop detection & watchdog for automatic error recovery
• Open completed tasks directly in the editor or LLM

Transcribe microphone or system audio in real-time while recording. No need to take notes during meetings and interviews — the text appears as you speak.

• Simultaneous recording + transcription — get both the audio file and text
• System audio capture (whole system / specific app) for real-time recognition (Pro)
• Level meter for real-time input monitoring

Automatically identify who said what in multi-speaker audio. Essential for creating meeting minutes and interview transcripts.

• High-accuracy speaker separation powered by a dedicated engine
• Auto-detect or manually specify speaker count (2–10) per file
• 24-color palette for visual speaker identification in the editor and audio bar
• Add, rename, delete speaker tags, and batch-change speakers across segments

A dedicated editing tool for efficiently finishing transcription results. Works with or without speaker diarization. Features per-segment audio playback, a fully keyboard-driven workflow, and auto-recovery. Dramatically speeds up proofreading for meeting minutes and interview records.

• Fully keyboard-driven (E=edit, D=split, M=merge, T=timing, S=speaker, Space=play, arrows=navigate)
• Auto-recovery: editing state is restored even after a crash or unexpected app close
• Speaker color-coded audio bar — click to seek, visually grasp the overall structure
• Variable speed playback (0.5x–2.0x) — speed audio files pre-generated in the background
• Multi-tab support — edit multiple files with independent state simultaneously
• Undo/redo for all 11 operation types (text edit, speaker change, split, merge, delete, insert, timing, etc.)
• Overlap detection during timestamp adjustment — auto-suggests resolution strategies

Capture audio via microphone, PC system audio, or YouTube/URL download. Recordings are automatically added to the transcription queue for a seamless record-to-text workflow.

• Auto-add recordings to the file list after completion
• Level meter and recording timer for real-time monitoring

Local AI chat running entirely on your machine. Load transcription results and ask "Summarize this" or "What are the key points?" — the AI analyzes the content locally. Confidential data stays safe.

• Streaming responses with real-time AI output
• Create, edit, and manage custom prompt templates
• Save, rename, restore, and export conversation history
• Adjust context size and control the LLM server from the GUI

Add subtitles to videos using transcription results. Supports both hardcoded (burned-in) and soft subtitles (as a track). Subtitle files can also be used for YouTube uploads.

• Hard sub: burned into the video (subtitles always visible)
• Soft sub: embedded as a track (can be toggled on/off)
• Customize font, size, color, and position

Pair with WhisperApp for Android (free) to record on your phone and leverage your PC's GPU for fast, high-accuracy transcription. Perfect for field recordings processed at your desk. (Currently in development — coming in a future update)

• Local Wi-Fi communication — no internet needed
• WebSocket for real-time progress updates on your phone
• Leverage PC GPU (CUDA/OpenVINO/Vulkan) from your smartphone

Freely choose and manage AI models for speech recognition and LLM. Beginners get recommended models, while advanced users can add any model from HuggingFace. Auto-detected GPU/VRAM info helps you check hardware compatibility before downloading.

• Recommended models: optimal quantization per model size, marked with a badge for easy selection
• Custom models: search HuggingFace or enter a direct URL to add any model (fine-tuned, etc.)
• GPU/VRAM auto-detection: hardware info automatically detected and displayed in status bar
• LLM models include VRAM/RAM requirements for easy compatibility checking
• 6 ASR models with quantization variants + 12+ LLM series — extensive model library
• Up to 3 concurrent downloads, with full download management and deletion

Check and install updates for transcription, LLM, audio processing, and other engines — all from within the app. The right build for your GPU is selected automatically.

• View all engine statuses from the Settings → Update tab
• One-click update or install for each engine
• Auto-check on startup with optional auto-install
• Auto-selects builds matching your GPU (CUDA/OpenVINO/Vulkan/CPU)

Automatically selects the optimal GPU backend for your hardware. Detects power source in real-time, balancing performance and battery life on laptops. Works perfectly without any configuration.

• Auto-detects NVIDIA GPU (CUDA), Intel GPU/NPU (OpenVINO), and Vulkan-compatible GPUs to select the optimal backend
• Real-time power source detection, automatically switching between performance and power-saving modes
• Automatic fallback to another backend on GPU errors — always stays stable
• Choose from 4 profiles: Performance / Balanced / Power Saving / Auto
• Manual backend selection also available for advanced users
• Each engine has different supported backends — the app automatically determines the best combination for each