FAAC Listening Tests: Compare Audio Quality

Find the AAC settings that sound right with your music. Start with a quick browser conversion, then use a controlled listening test when you need a reliable comparison.

Looking for implementation support or speed measurements? Use the selection guide or benchmark results.

Try your own music

The browser converter lets you encode a file locally and download an M4A. Start with a lossless source such as WAV or FLAC supported by your browser. Reencoding an MP3 or another lossy file adds a second generation of loss and complicates the comparison.

The demo is useful for exploration. For a reproducible comparison, use a local CLI build whose version and settings you can record. The browser demo uses its own compiled encoder and browser audio processing; a local build gives you more control over the test.

Make a controlled comparison

1. Keep the source fixed. Use the same lossless passage for each encoder. Include attacks, cymbals, sustained tones, vocals, and stereo material you find challenging.
2. Match the settings. Record sample rate, channels, AAC profile, rate-control mode, target bitrate, and actual output bitrate. Compare AAC-LC with AAC-LC and HE-AAC v1 with HE-AAC v1.
3. Decode consistently. Use the same decoder to create PCM files, align them with the source, and check for level differences. Avoid playback effects or normalization that changes only one candidate.
4. Listen without knowing the identity. Use an ABX tool to compare the reference with an encoded version. To compare preference between encoders, use a blind randomized comparison with the reference available. Record trial counts and results, not just a sighted impression.
5. Retain the evidence. Save commands, encoder and decoder versions, source details, and the specific time range where an artifact occurs.

AAC is lossy. Keep a lossless master for archiving, even if an AAC copy sounds transparent to you.

Report a quality issue

Share reproducible quality findings in FAAC issues.

Include:

• Encoder version or commit, build options, and operating system.
• Exact command, profile, sample rate, channels, and measured bitrate.
• Decoder version and listening-test method, including trial results when available.
• A short source sample you have permission to share, or instructions for reproducing it.
• The affected time range and what you hear, such as pre-echo, tonal noise, or a changed stereo image.

Specific findings help improve the encoder. Reports from experienced listeners, including people disappointed by older FAAC versions, are welcome.