WAV vs FLAC: How to Check Whether Smaller Files Contain the Same Audio

You have a finished audio file to store or hand over. FLAC can be smaller than WAV, but file size alone cannot tell you whether the exported audio stayed the same. A useful check compares the decoded samples in a common representation.

I'm Alex from the Creatune team. I used the logged-out Creatune Audio Converter to export the same self-created six-second test tone as WAV and FLAC, then inspected both files with FFmpeg. This is a file and sample comparison, not a listening test or a claim about every possible input.

What the two formats store

A WAV container commonly carries uncompressed PCM audio. FLAC compresses audio losslessly: decoding a correctly encoded FLAC stream recovers the samples represented by that stream. That does not establish that an entire conversion pipeline preserved the source's original channel count, sample rate, or metadata.

For a handover, define what must match before comparing files:

Required: same duration, sample rate, channels, and decoded samples
Separately check: filename, tags, artwork, and receiving-app compatibility
Keep: the original source

Export the same source twice

My source was a six-second, mono, 44.1 kHz, 16-bit PCM WAV. A simple tone is useful for a controlled comparison, although its compression ratio will not represent a dense music mix.

Open the converter and choose a local audio file you can use.
Wait for the waveform and duration to appear.
Open Output format, choose WAV, and download it.
Keep the same loaded source, switch Output format to FLAC, and download again.

The page lists MP3, WAV, and FLAC as output choices. It describes local editing as free without generation credits and warns that edits are not saved after leaving the page.

Name the downloads converted.wav and converted.flac before running the commands below.

Inspect the output properties first

ffprobe -v error -select_streams a:0   -show_entries stream=codec_name,sample_rate,channels,duration   -of json converted.wav

ffprobe -v error -select_streams a:0   -show_entries stream=codec_name,sample_rate,channels,duration   -of json converted.flac

Actual results from this run:

Property	WAV	FLAC
File bytes	1,058,444	312,116
Duration	6.000000 s	6.000000 s
Sample rate	44,100 Hz	44,100 Hz
Channels	2	2
Codec	pcm_s16le	flac

Both exports were stereo even though the source was mono. Therefore, this test compares the two exports; it does not demonstrate source-format preservation. That distinction matters if a recipient specifically needs mono delivery.

Compare the decoded samples

FFmpeg is required for this example. Decode both outputs into the same signed 16-bit little-endian representation, without forcing a new sample rate or channel count:

ffmpeg -v error -i converted.wav -map 0:a:0   -c:a pcm_s16le -f s16le wav.pcm
ffmpeg -v error -i converted.flac -map 0:a:0   -c:a pcm_s16le -f s16le flac.pcm

Then use Python's standard library to compare them:

from pathlib import Path
import hashlib

wav = Path("wav.pcm").read_bytes()
flac = Path("flac.pcm").read_bytes()
print(len(wav), len(flac))
print(wav == flac)
print(hashlib.sha256(wav).hexdigest())
print(hashlib.sha256(flac).hexdigest())

The actual output length was 1,058,400 bytes for each decoded file. The sample bytes matched, and both SHA-256 hashes were:

3f1cf67d5bf71afc47a0befa8f6a9e7722fba804ad92ce5cdeca05afce755628

This establishes equality in the tested 16-bit decoded representation. For higher-bit-depth material, choose an appropriate comparison representation rather than silently discarding precision.

Use the result in a delivery decision

For this particular tone, FLAC used about 29.5% of the WAV file's storage while decoding to the same tested sample bytes. Real music, silence, and noise compress differently. Do not turn that percentage into a general promise.

Use WAV when the receiving workflow asks for it; use FLAC when lossless compressed storage and application support suit your needs. Check tags and artwork separately because decoded PCM equality says nothing about metadata. Keep the source, especially when the output channel configuration differs.