Orbit Documentation Français

Audio and transcription

Audio Studio

Processes the audio of a sound file or of a video.

The Audio Studio
The Audio Studio
  1. Drop your audio files or videos in the area (several at once).
  2. Select a file and pick the Processing:
Processing What it does
Stem separation splits vocals from music: 2 stems (vocals + accompaniment), 4 stems (vocals, drums, bass, other) or, with Demucs, 6 stems (+ guitar and piano). Premium.
Denoise reduces hiss and constant background noise; set the strength.
Normalize brings the volume to the level of streaming platforms (-14 LUFS, peak -1.5 dBTP).
Extract audio takes the sound track out of a video, without re-encoding the picture, as WAV, MP3 or FLAC.
  1. Apply to the whole queue if needed, then Run queue.

For separation, two engines:

  • High quality (Demucs v4): much better; with an NVIDIA card, it uses the GPU. Install: Demucs + PyTorch GPU (about 2.5 GB) or CPU (about 700 MB). Stems come out as 320 kb/s MP3. Fast (about 4 times faster) or Max quality (fine-tuned model).
  • Fast & light (Spleeter): fair quality, about 700 MB to install.

Transcription

Turns the speech of a video or audio file into text and subtitles, on your computer. Part of Orbit Premium.

Transcription
Transcription
  1. Choose a video or audio (MP4, MKV, MOV, MP3, WAV…).
  2. Language: Auto-detect or the spoken language.
  3. AI model: from the fastest (Base, about 142 MB) to the most accurate (heavier and slower).
  4. Export to: tick the formats you want. Orbit prepares the right file for each program: Premiere Pro, After Effects, CapCut, DaVinci Resolve, Final Cut Pro, YouTube / Web, Styled subtitles, Plain text, Data and Lyrics.
  5. Burn subtitles into the video also creates an .mp4 with burned-in subtitles (not editable); set the font size and color.
  6. Output folder (next to the source file by default), then transcribe and export.

When it is done, Open shows the files created.