How to compress audio for transcription

Method 1 — Extract audio from a video file

If you have a video file (MP4, WEBM, MOV) that's mostly speech, the audio track is usually 5–10× smaller than the video. Mictoo only needs the audio — the video doesn't help transcription. This is the fastest fix for most users.

Online (no install)

Use our own browser converter — drop the file, get MP3 back in a few seconds. Free, no signup, no watermark:

• MP4 to MP3 — pulls the audio out of any MP4 video.
• WEBM to MP3 — for YouTube downloads and screen recordings.
• WAV to MP3 — shrinks an uncompressed WAV to a manageable MP3.

Command line (ffmpeg)

ffmpeg -i input.mp4 -vn -acodec copy output.m4a

-vn strips video, -acodec copy keeps the audio quality intact. Fast — no re-encoding.

Method 2 — Lower the audio bitrate

Most podcasts and meetings are recorded at 128–192 kbps. For transcription, 64 kbps mono is more than enough — accuracy stays the same and the file becomes 2–3× smaller. This is the best method for long audio files.

Online

We don't host a re-encode tool yet, so for bitrate reduction the easiest path is the ffmpeg one-liner below. If you prefer a GUI, Audacity (free, all platforms) handles it via File → Export → Export as MP3 with the quality slider set to 64–96 kbps.

Command line (ffmpeg)

ffmpeg -i input.mp3 -ac 1 -b:a 64k output.mp3

-ac 1 converts to mono (speech is fine in mono),-b:a 64k sets audio bitrate to 64 kbps.

Rough math: 64 kbps gives ~28 MB per hour. A 1-hour interview fits in roughly 28 MB — close to the limit. Drop to -b:a 48k for ~21 MB/hour without noticeable accuracy loss.

Method 3 — Split into shorter parts

If you need to keep the original quality (legal, medical, archival), split the file into chunks of 20–25 minutes and transcribe each separately. Then join the transcripts in any text editor.

Online

Splitting in the browser needs more horsepower than our serverless converter has today. For now the ffmpeg one-liner below is the cleanest route, or open the file in Audacity and export each region separately. See the dedicated how to split a long audio file guide for both methods step by step.

Command line (ffmpeg)

ffmpeg -i input.mp3 -f segment -segment_time 1200 -c copy part_%03d.mp3

Splits the input into 20-minute (1200 seconds) parts named part_001.mp3, part_002.mp3, etc. Then upload each to Mictoo.