Audio Import WAV · MP3 · M4A · FLAC · WEBM

How to transcribe audio files locally on your computer

Import existing interviews, Android voice memos, or legacy recordings (up to 256 MiB and 3 hours) into Stellar Recorder while preserving your original source file untouched.

Local Audio Import & Transcription Pipeline
Audio Import WAV · MP3 · M4A · FLAC · WEBM
01 · Validate

Check Format & Size

Supports WAV, MP3, M4A, FLAC, OGG, AAC, OPUS, WMA and WEBM up to 256 MiB / 3h.

02 · Import

Assign Client Project

Choose target project, enter optional original date, and leave original source untouched.

03 · Process

Run Speech & Diarization

Transcribes first audio track as one mixed source; Pro/trial can separate voices.

04 · Inspect

Search & Name Speakers

Review transcript timestamps, rename speakers per file and check names/numbers.

Stellar Recorder - Import Audio File Dialog (WAV / MP3 / M4A)
WAV · MP3 · M4A · FLAC Import
Import audio modal showing attached M4A recording, target project selector, and process toggle

Prepare your audio file cleanly

Verify your file format is WAV, MP3, M4A, FLAC, OGG, AAC, OPUS, WMA or WEBM under 256 MiB and 3 hours duration. Only the first audio stream is processed, your source file remains untouched, and recording consent remains your responsibility regardless of file origin.

Dates, single mixed sources & speaker labels

Imported files transcribe as one combined audio track rather than separate mic and loopback streams. If original recording metadata is unknown, the reader marks the original date as unknown and shows import time separately. Pro or 14-day trial installs can split voices locally via Nemotron3Q8, though manual review against audio remains essential.

Import and transcribe step by step

  1. Select Import audio inside Stellar Recorder and pick your supported file.
  2. Choose the destination project folder (selection locks while processing runs).
  3. Enter the original recording date if known for accurate chronological sorting.
  4. Confirm your speech model (the Faster 25-language model or the Slower, More Precise 100-language model) and start local transcription.
  5. Inspect finished timestamps, search keywords and verify names or numbers in the reader.