How to extract audio from a video file (MP3, M4A, WAV in 2026)

Extracting audio from a video file is one of those operations that should be a single click but often turns into a ffmpeg search-and-paste session. The actual mechanics are simple. The choice that matters is which output format you pick, and whether you can skip re-encoding entirely.

This guide is the practical version: when to stream-copy without re-encoding (the lossless 3-second path), when to re-encode for compatibility, and the click path in Video Forge’s Smart Actions panel.

TL;DR

For most cases, the right output format is M4A (AAC audio in an MP4 container) because that’s the codec already inside almost every modern video file — no re-encoding needed, output is bit-identical, takes about 1 second per gigabyte.

Three formats, when each wins:

Format	When to use	Operation
M4A (AAC)	Default. Modern source, modern playback.	Stream-copy, lossless, fast.
MP3	Older devices, audio for embedding in podcasts, MP3-only systems.	Re-encode, small quality loss.
WAV	Feeding into Audacity, Logic, Premiere; archive; broadcast.	Stream-copy if source is PCM, else re-encode.

In Video Forge: Tools menu → Extract audio → pick MP3 / M4A / WAV → Convert. The app reads the source’s audio codec and picks stream-copy when possible. Output lands next to the source with _audio.m4a (or .mp3 / .wav).

What’s actually in the audio track

When you record video on any modern device — iPhone, Android, DSLR, screen recording, OBS — the audio inside the video file is one of:

Source	Audio codec	Right output
iPhone video	AAC	M4A (stream-copy)
Android video	AAC	M4A (stream-copy)
Mac screen recording	AAC	M4A (stream-copy)
OBS screen capture (default)	AAC	M4A (stream-copy)
DSLR / mirrorless	PCM or AAC	WAV (if PCM) or M4A
ProRes intermediate	PCM	WAV (stream-copy)
YouTube download (mp4)	AAC	M4A (stream-copy)
Old AVI from 2007	MP3	MP3 (stream-copy)

For ~95% of consumer video, the audio is already AAC. Wrapping the existing AAC stream into an M4A container is a remux — no re-encoding, no quality loss, completes in about a second.

If you specifically need MP3 (older playback compatibility, hosting on a service that only accepts MP3, embedding in a podcast feed that prefers MP3), the AAC has to be re-encoded. That’s a quality hit, but at 192 kbps or higher it’s usually inaudible.

MP3 vs M4A vs WAV

Each format trades off in a specific direction. The honest breakdown:

M4A (AAC in MP4 container)

• What it is: the same codec used inside every modern MP4 video, wrapped in a container designed for audio.
• Quality at given bitrate: best of the three at lossy bitrates. AAC at 128 kbps is roughly equivalent to MP3 at 192 kbps.
• Compatibility: every Apple device, Android, modern Windows, modern browser. The one weak spot is old hardware MP3 players from before ~2010.
• Stream-copy from video: yes, almost always, because most video already contains AAC audio.
• File size: about 30% smaller than MP3 at equivalent quality.

The right default for almost every audio-extract operation.

MP3 (MPEG-1 Audio Layer III)

• What it is: the original lossy audio compression format, standardized in 1993.
• Quality at given bitrate: worse than AAC at the same kbps. The difference becomes noticeable below 128 kbps.
• Compatibility: literally every audio playback device ever made. This is its actual selling point — universality.
• Stream-copy from video: rarely, because video files almost never contain MP3 audio in 2026.
• File size: ~30% larger than M4A at equivalent quality.

Use MP3 when:

• The destination is a hardware MP3 player, car stereo, or other device older than ~2010.
• The destination platform specifically requires MP3 (some podcast hosts, some legacy upload forms).
• You’re sharing with someone who explicitly asked for MP3.

WAV (uncompressed PCM)

• What it is: raw uncompressed audio samples. No compression, no lossy quality hit.
• Quality: perfect (it’s the source, basically).
• Compatibility: every playback device and every editing tool.
• Stream-copy from video: yes if the source is PCM (rare in consumer video, common in DSLR and ProRes).
• File size: ~10× larger than M4A. A 5-minute stereo 48 kHz WAV is about 55 MB.

Use WAV when:

• Feeding audio into editing software (Audacity, Logic, Premiere, Resolve) — they don’t care about file size, they care about not re-decoding lossy audio.
• Archive or broadcast spec.
• The source itself is PCM and you want lossless extraction.

For anything else, M4A or MP3.

The stream-copy shortcut

This is the part most online tutorials miss: if you don’t need to change the audio codec, ffmpeg’s -c:a copy flag pulls the existing audio stream straight into the new container without re-encoding.

# Source is MP4 with AAC audio (the common case). Output: M4A, lossless.
ffmpeg -i input.mp4 -vn -c:a copy output.m4a

The -vn flag means “no video” — drops the video stream entirely. The -c:a copy means “copy the audio stream as-is.” Combined, this is the fastest possible audio extraction and the result is bit-identical to what was in the source video.

On a typical 1 GB MP4, this takes about 1 second. The output file is the size of the audio portion only — typically 5-20 MB depending on duration and bitrate.

If your source is PCM audio (DSLR or ProRes), the same trick works to WAV:

ffmpeg -i input.mov -vn -c:a copy output.wav

If you actually need to change the codec — most often, “I want MP3”:

ffmpeg -i input.mp4 -vn -c:a libmp3lame -b:a 192k output.mp3

That’s a re-encode, so it’s slower (takes a few seconds for a long video) and lossy. 192 kbps is the sweet spot for MP3 — high enough that the loss is inaudible to most listeners, low enough that file size stays reasonable.

The step-by-step in Video Forge

Audio extraction is a smart action in Video Forge — it runs outside the main job queue because it’s a single fast operation:

1. Open Video Forge.
2. Drop a video onto the queue, or have one already loaded.
3. Open Tools → Extract Audio.
4. Pick the output format — MP3, M4A, or WAV.
5. Click Extract. The action reads the source’s audio codec and picks stream-copy when possible. The output lands next to the source with the suffix _audio.m4a (or the corresponding format).

A completion toast appears when the file’s ready. Click the path in the toast to reveal the file in Finder / Explorer.

For batch extraction (Library mode → select multiple → Extract Audio), the operation runs per file in parallel. Useful for converting a folder of video files into a podcast-ready audio archive.

When stream-copy fails

A few edge cases where -c:a copy won’t work and you have to re-encode:

1. Source audio is in a codec the output container doesn’t support. AC-3 audio in MP4: technically allowed but flaky in some players. Best to transcode to AAC for M4A output.
2. Source has multiple audio tracks (foreign-language dubs, commentary tracks). -map 0:a:0 picks the first audio track explicitly.
3. Source audio uses an unusual sample rate or channel layout that the destination format doesn’t accept. Rare in practice but happens with 96 kHz studio captures going to MP3 (MP3 maxes at 48 kHz).

Video Forge handles all three automatically — if stream-copy isn’t viable, it falls back to a re-encode using sensible defaults (AAC 192 kbps for M4A, libmp3lame 192 kbps for MP3, pcm_s16le for WAV).

Use cases this fits

• Podcast production — pull the audio from a Zoom or Riverside recording, drop into editing software. M4A or WAV.
• Music from a music-video file — extract the audio for a personal audio library. M4A (lossless from the source AAC) or MP3 (for older device compatibility).
• Transcription prep — audio-only file feeds into Whisper, Otter, etc. M4A is the right format; most transcription tools accept it.
• Dialogue cleanup — pull audio from raw video, run noise reduction in Audacity, then put it back. WAV both directions.
• Vocal-only YouTube downloads — audio-only for radio-style listening on a phone, saves data. M4A.

For “I want to put the cleaned audio back into the video,” see VideoForge’s audio replace flow (planned feature).

Speed comparison

Tested on a 1 GB MP4 (10 min, 1080p, AAC stereo at 192 kbps), M2 Pro, ffmpeg 7.0:

Operation	Time	Notes
Stream-copy to M4A	0.8s	Disk-IO bound
Re-encode to MP3 (192 kbps)	4.5s	libmp3lame
Re-encode to WAV	1.2s	Just decoding, no encoding cost
Full video → audio CLI search-and-paste	~3 min	Realistic when you don’t know the flags

The 3-minute number isn’t a joke. The reason audio extraction tutorials exist is that finding the right ffmpeg flags is harder than running them.

FAQ

How do I extract audio from a video file? Easiest: stream-copy the existing audio into an M4A container. Lossless, ~1 second per gigabyte. ffmpeg -i input.mp4 -vn -c:a copy output.m4a or use Video Forge’s Tools → Extract Audio panel.

What’s the difference between MP3, M4A, and WAV? MP3: universal compatibility, lossy. M4A: best quality at lossy bitrates, smaller than MP3, modern. WAV: uncompressed, perfect quality, large file. M4A is the right default; MP3 for older device support; WAV for editing software.

Does extracting audio reduce quality? Stream-copy extraction is lossless (output bit-identical to source). Re-encoding extraction loses a small amount of quality, usually inaudible at 192 kbps and up. WAV extraction is always lossless.

Can I extract audio from a YouTube or streaming video? Not directly through a desktop converter — download first (yt-dlp is the standard tool), then extract from the local file. Be aware of terms of service.

Does Video Forge handle multi-file audio extraction? Yes — switch to Library mode, multi-select, run Tools → Extract Audio. Each file is processed in parallel.

---

Video Forge does audio extraction as a smart action — pick MP3, M4A, or WAV, click Extract, get a file in seconds. 10 conversions are free on macOS and Windows; ffmpeg is bundled. See also the GIF-from-video guide for another smart action, or convert MOV to MP4 for the lossless video-only sibling operation.