DaVinci ASR is a local AI subtitle generation tool built for DaVinci Resolve, powered by Alibaba Qwen's Qwen3 ASR models.

Choose Qwen3-ASR-0.6B or Qwen3-ASR-1.7B for speech recognition, paired with Qwen3-ForcedAligner-0.6B for timestamp alignment. Turn video or audio from your DaVinci Resolve timeline directly into subtitles.

After the models are downloaded, speech recognition, alignment, and subtitle generation can all run locally on your computer.

Available for macOS and Windows, with a choice of Qwen3-ASR-0.6B or Qwen3-ASR-1.7B.

What Problem Does It Solve?

Creating subtitles for interviews, courses, podcasts, documentaries, and long-form videos often involves several separate steps: exporting audio, transcribing it in another application, exporting an SRT file, and importing it back into DaVinci Resolve.

DaVinci ASR brings these steps into one workflow inside DaVinci Resolve.

It can process audio from the current timeline, transcribe it with Qwen3 ASR, align the recognized text with the original audio using Qwen3 Forced Aligner, generate subtitle timing, and import the resulting subtitles back into DaVinci Resolve.

The core subtitle generation process runs locally, making it especially useful for long-form content, private footage, and projects with large amounts of dialogue.

Core Features

• Two Qwen3 ASR Model Options

  Choose Qwen3-ASR-0.6B or Qwen3-ASR-1.7B for local speech recognition.

  Once the model is installed, transcription runs directly on your computer without uploading project audio to a cloud ASR service.

• Qwen3 Forced Aligner for Accurate Timing

  Uses Qwen3-ForcedAligner-0.6B to align recognized text with the original audio and generate accurate subtitle start and end times.

  Speech recognition and timestamp alignment are handled by dedicated Qwen3 models.

• Support for 11 Subtitle Languages

  Currently supported languages include:

  Chinese Mandarin, English, Cantonese, French, German, Italian, Japanese, Korean, Portuguese, Russian, and Spanish.

  Automatic language detection is also available.

• Custom Prompt for Names and Terminology

  Add names, brands, product names, technical terms, and contextual information to help Qwen3 ASR better understand project-specific vocabulary.

• Automatic Voice Activity Detection

  Built-in Silero VAD detects speech regions on the original timeline and works together with Qwen3 ASR for segmented transcription.

  Long audio is processed in chunks to reduce memory usage during transcription.

• Subtitle Length Control

  Set the maximum number of characters per subtitle.

  DaVinci ASR uses punctuation, meaningful pauses, and word boundaries when planning subtitle segments.

  CJK languages automatically use a more suitable character limit for readability.

• Reduce Subtitle Gaps

  Enable the No Gaps option to reduce unnecessary empty spaces between consecutive subtitles and create a more continuous subtitle track.

• Remove End Punctuation

  End punctuation can be removed automatically for subtitle styles commonly used in short-form videos and social media content.

• Built-in Subtitle Editor

  Review subtitle index numbers, start times, end times, and subtitle text directly inside the tool.

  Subtitle text can be edited and updated after generation.

• Find and Replace

  Built-in text search, Replace, and Replace All functions make it easier to correct names, terminology, and repeated recognition errors.

• Integrated DaVinci Resolve Workflow

  DaVinci ASR can generate audio from the Resolve timeline, run Qwen3 ASR transcription and alignment, create an SRT file, and automatically import it back into the current project.

  The full workflow stays inside the DaVinci Resolve editing environment.

Qwen3 Local Models

Choose either speech recognition model in the tool:

• Qwen3-ASR-0.6B
• Qwen3-ASR-1.7B

The selected ASR model transcribes the audio. Qwen3-ForcedAligner-0.6B then aligns the text with the original audio to generate subtitle start and end times.

Download models directly from the tool. Users in mainland China can choose ModelScope; users in other regions can use Hugging Face.

Downloaded models are stored locally and do not need to be downloaded again for future transcription tasks.

Compatibility

• Operating systems: macOS, Windows
• Editing software: DaVinci Resolve

Install the version for your operating system. After downloading the models, speech recognition, timestamp alignment, and subtitle generation run locally on your computer.

On Macs with Apple MPS support, the runtime can use Apple GPU acceleration. If MPS is unavailable, it falls back to CPU processing.

DaVinci ASR includes its own runtime environment. The standard release does not require a separate installation of Python, pip, Conda, or Homebrew.

Workflow

1. Launch DaVinci Resolve on macOS or Windows.
2. Open the project and timeline you want to subtitle.
3. Open DaVinci ASR.
4. Choose Qwen3-ASR-0.6B or Qwen3-ASR-1.7B, then download the selected model and Qwen3 Forced Aligner on first use.
5. Select the language of your video or use automatic language detection.
6. Add names, brands, terminology, or contextual prompts if needed.
7. Set the maximum characters per subtitle.
8. Optionally enable No Gaps or Remove End Punctuation.
9. Click Create Subtitles.
10. DaVinci ASR processes the timeline audio and runs Qwen3 ASR.
11. Qwen3 Forced Aligner aligns the transcription with the original audio.
12. Subtitles are generated and imported into DaVinci Resolve.
13. Review, edit, search, or replace subtitle text in the built-in subtitle editor.

Use Cases

DaVinci ASR is suitable for:

• Long-form interviews and podcasts
• Courses, tutorials, and educational videos
• Documentaries and interview-based content
• YouTube videos
• Short-form and social media videos
• Multilingual projects including Chinese, English, Cantonese, and Japanese
• Videos containing many names and technical terms
• Projects that require local audio processing
• Fast subtitle draft generation
• macOS and Windows users who want to create subtitles directly inside DaVinci Resolve

Frequently Asked Questions

Which Qwen3 ASR models can I choose?

Choose Qwen3-ASR-0.6B or Qwen3-ASR-1.7B for speech recognition. Qwen3-ForcedAligner-0.6B handles subtitle timing alignment.

The selected ASR model and the alignment model run through the local DaVinci ASR runtime.

Which languages are supported?

The current version supports timestamped subtitles for:

Chinese Mandarin, English, Cantonese, French, German, Italian, Japanese, Korean, Portuguese, Russian, and Spanish.

Automatic language detection is also supported.

Can it run completely locally?

Yes.

After the Qwen3 models have been downloaded, speech recognition, timestamp alignment, and subtitle generation can run locally on your computer.

An internet connection is required when downloading the models for the first time.

Does it support Windows?

Yes. DaVinci ASR supports macOS and Windows. Both versions run transcription, timestamp alignment, and subtitle generation locally.

Do I need to install Python?

No.

The standard release includes its own runtime environment and does not require a separate installation of Python, pip, Conda, or Homebrew.

Can I add names and technical terms?

Yes.

The Prompt field can include names, brands, product names, technical terminology, and contextual information to help Qwen3 ASR recognize project-specific vocabulary.

Can I control subtitle length?

Yes.

You can configure the maximum number of characters per subtitle. DaVinci ASR uses punctuation, semantic pauses, and word boundaries when generating subtitle segments.

Can I edit generated subtitles?

Yes.

DaVinci ASR includes a built-in subtitle editor with text editing, search, Replace, and Replace All functions.

Does it support speaker recognition?

Speaker diarization and speaker identification are not included in the current version.

Does it use Whisper?

DaVinci ASR uses Qwen3 ASR. Choose Qwen3-ASR-0.6B or Qwen3-ASR-1.7B for transcription, with Qwen3-ForcedAligner-0.6B for timestamp alignment.

Buy DaVinci ASR

Check the price at the top of the page and purchase DaVinci ASR directly.

Use Qwen3 ASR inside DaVinci Resolve on macOS or Windows for local speech recognition, timestamp alignment, automatic subtitle generation, and subtitle editing.