DubDuck vs Linly-Dubbing

DubDuck is a web service for translating and dubbing video dialogue. It transcribes and translates speech, clones speakers’ voices, and lip-syncs dubbed video into 40+ languages. Its editor provides dialogue and translation workspaces, speaker controls, a frame-accurate timeline, and segment-level re-dubbing. The documented workflow covers upload, source and target language selection, transcription and editing, calibration, dubbing, preview, and export. New accounts start with $10 in free credits, while videos dubbed before payment show a visible DubDuck.com watermark. Beta documentation limits uploads to 3.5 minutes and 50 MB, although the homepage describes uploads as any format and length. Dubbing costs $1.95 / min per minute of dubbed dialogue, billed by the second; only dialogue is charged, and calibration is billed the same way during the first dub. Top-ups cost $5.00 per once, $10.00 per once, or $25.00 per once. Credits do not expire while the account remains active. Voice characteristics are used for the project and discarded shortly after rendering.

Linly-Dubbing is an open-source tool for translating and dubbing videos into Chinese and other languages. It combines speech recognition and speaker recognition with translation options using OpenAI models, locally run Qwen, or Google Translate. Documented voice synthesis choices include Edge TTS, XTTS, CosyVoice, and GPT-SoVITS; voice cloning can generate speech resembling the original video's voice. Linly-Talker lip-sync technology aligns dubbed speech with the video. The tool can add subtitles and background music, adjust music volume, change playback speed, separate vocals from accompaniment, and download video and audio through yt-dlp. Setup instructions describe a local WebUI and use of Python 3.10, PyTorch 2.3.1, and CUDA 11.8 or 12.1, with Windows and Linux model-download instructions. A Colab script is also linked for online use. Installation requires configuring dependencies, downloading models, and setting environment variables, and may use external service credentials depending on configuration. The project is licensed under Apache License 2.0; its releases page lists no releases.