What exactly is transcoding loss?

It is the cumulative degradation that occurs when you decode a lossy file and re-encode it lossily. When your MP3 was created, its psychoacoustic model permanently discarded a portion of the audio information. Converting to AAC decodes that already-incomplete audio and a second encoder discards information again according to its own rules. Because the two encoders use different models, the losses do not overlap neatly — they accumulate.

Isn't AAC a better codec? Why doesn't that help?

AAC genuinely is more efficient than MP3 at the same bitrate, but that advantage only applies when encoding from a lossless source. A codec's efficiency is about how well it preserves the information it is given, not about producing information that does not exist. Frequency data deleted by MP3 never reaches the AAC encoder at all, so AAC cannot rescue it.

Is the loss actually audible?

It depends on the source bitrate and the content. Converting a 256-320 kbps MP3 to 256 kbps AAC, most listeners will not hear a difference on most equipment. But convert a 128 kbps source to a low AAC bitrate and you will notice a metallic character or softening on cymbals, applause and hi-hats. The lower the source bitrate, the more audible the transcoding loss becomes.

Is there any way to avoid the loss entirely?

If you are starting from a lossy source, no, it cannot be avoided completely. But it can be seriously limited by three things: keeping the target bitrate close to or slightly above the source, going to your target in a single hop without passing through intermediate formats, and where possible not starting from MP3 at all but encoding from a lossless source such as FLAC or WAV. If the third option is available, the other two become unnecessary.

Transcoding Loss in MP3 to AAC Conversion: Why Quality Never Improves

There is a piece of misinformation that circulates widely online: "improve your audio quality by converting your MP3 files to AAC." The claim starts from the genuine fact that AAC is a better codec, but arrives at the wrong conclusion. This article explains how double compression works, why a quality gain is technically impossible, and when the conversion still makes sense despite that.

What does lossy compression delete?

MP3 and AAC are both lossy codecs. Both use the same fundamental idea: exploit the characteristics of human hearing to discard audio information that is unlikely to be perceived.

The mechanism that does this is called a psychoacoustic model, and it rests on two phenomena:

Masking: Quieter frequencies sitting immediately next to a loud one are not perceived by the ear. The encoder sees no reason to store them and throws them away.
Threshold of hearing: Very high frequencies (particularly above 16 kHz) and very low-level sounds are inaudible anyway. These are discarded as well.

This discarding is irreversible. The information is simply not in the file — it is not sitting there compressed, waiting to reappear when decoded. That is precisely what distinguishes it from a ZIP file.

Two encoders, two different models

The crucial detail is this: MP3 and AAC use different psychoacoustic models. Their decisions about what to discard do not line up exactly.

When you convert your MP3 to AAC, the following happens:

Decoding: The MP3 data is turned back into a raw waveform. But that waveform is not the original recording — everything discarded during MP3 encoding is missing. On top of that, small artefacts generated by MP3 encoding itself are baked into this waveform.
Re-encoding: That incomplete waveform is handed to the AAC encoder. And here is the critical problem: the AAC encoder has no idea the audio has already been processed once. It treats it as an original recording and prunes it according to its own model.

The result is damage in two directions. On one hand, some details MP3 chose to preserve may be discarded under AAC's model. On the other — the more insidious effect — artefacts left behind by MP3 encoding are mistaken for real audio by AAC, and bitrate is spent preserving them. Part of your budget goes toward storing distortion.

This cumulative effect is called transcoding loss.

Why doesn't the "AAC is a better codec" argument apply here?

AAC's advantage over MP3 is real: thanks to more flexible block sizes, more advanced stereo coding and wider frequency support, it produces less audible distortion at the same bitrate. The gap is especially pronounced in the 96-128 kbps region.

But that advantage is about how well the encoder preserves the information it is given. The quality of an encoder does not translate into an ability to generate information that does not exist.

An analogy: you have a low-resolution, compressed photograph and you save it into a more capable image format. The new format may be technically superior, but it will not restore detail that is not in the photograph. It will simply store what you have with slightly more fidelity.

The same logic holds here: AAC never sees the frequency information that MP3 deleted. Recovering it is physically impossible.

What determines the size of the loss?

Transcoding loss is not always equally audible. The determining factors:

Source bitrate. The most important one. A 320 kbps MP3 carries plenty of information; convert it to 256 kbps AAC and the second encoder has little left to throw away, so the result is indistinguishable in most conditions. A 128 kbps source, by contrast, is already at the edge; adding a second round of pruning pushes it past the threshold quickly.

Target bitrate. Choosing a target markedly below the source magnifies the loss. A target close to or slightly above the source gives the second encoder breathing room.

Content type. Speech, podcasts and audiobooks are narrowband content and survive transcoding well. Cymbals, applause, hi-hats, string instruments and the high-frequency layers of electronic music are the most fragile regions; degradation appears here first, as a metallic or "watery" character.

Listening conditions. In a car, through a phone speaker or in a noisy environment, transcoding loss is very hard to hear. On good headphones in a quiet room it becomes noticeable.

So is the conversion ever worthwhile?

It is — but not for quality reasons. Valid motivations include:

Ecosystem fit. If you live on Apple devices and want your library consolidated into one format, AAC/M4A is the natural choice. Management, syncing and tagging all become simpler.

Storage savings. Because AAC is more efficient, stepping down to a slightly lower bitrate costs less quality than the equivalent step in MP3. If space is tight on your phone, that is a tangible advantage.

Platform requirements. Some broadcast infrastructures, video editing workflows and app development tools expect AAC/M4A input. If there is no alternative, you have to convert.

Library consolidation. Having three different formats across an archive of thousands of tracks complicates backups and syncing.

If one of these applies, the conversion is a reasonable trade-off. If none of them does and your MP3 already plays fine, leaving it alone is the best decision.

Three rules for minimising loss

1. Look for a lossless source. If a FLAC or WAV version of the same content exists, produce the AAC directly from it. One encoding pass always beats two, and only this way can you benefit from AAC's genuine efficiency.

2. Keep the target bitrate close to the source. 192 kbps AAC for a 192 kbps MP3, 256 kbps AAC for a 320 kbps MP3 are reasonable targets. Going far above the source bloats the file without adding quality — because there is no information to add. Our MP3 to AAC converter lets you pick the bitrate yourself.

3. Do not chain conversions. A path like MP3 → AAC → OGG → MP3 adds new loss at each step. Go to your target format in a single hop, and do not delete the original — if you need a different setting later, start again from there.

In summary

Converting MP3 to AAC does not improve your audio; it cannot. Information deleted during lossy encoding is permanently gone and no codec can bring it back. The genuine reasons for this conversion are ecosystem fit, storage efficiency or platform requirements. If one of those applies to you, choosing a bitrate close to the source and converting in a single pass will keep the additional loss inaudible in most use. If none applies, the best strategy is not to convert at all.