MP3 audio quality is decided by four things in this order: the source recording, the encoder, the bitrate and the equipment you play the file on. The bitrate printed on a file is only one of the four. The weakest of them sets the result, and no later step can make up for an earlier one.
This article follows the four in that order. For the format itself, its frames and its tags, start with what an MP3 file is.
The source sets the ceiling
An MP3 encoder can only remove information. Whatever the source lacks, the MP3 lacks too. A file encoded from a CD or a lossless download starts from the full recording. A file made from an online video does not. YouTube stores audio as AAC or Opus at roughly 128 to 160 kbps for most videos, so an MP3 converted from it has been through lossy compression twice.
The recording matters as much as the format. A live clip filmed on a phone, a worn cassette transfer and a studio master are different sources, even when all three end up as 192 kbps files. When you search on MP3Juice, an official upload from the artist or label is usually a better starting point than a re-upload of unknown origin. Press Play to listen before you download.
The encoder: same bitrate, different results
The MP3 standard defines exactly how a file must be decoded. It leaves the encoder free to decide what to keep. Two programs given the same song and the same bitrate will therefore produce different files, and one may handle cymbals or applause far better than the other. What an encoder removes, and in which order, is set out in how MP3 compression works.
Encoders went on improving for years after the format appeared. A file made with a mature encoder such as LAME can sound cleaner than a file from the early 2000s at a higher bitrate. Settings matter as well, since the same encoder can be run in a quick mode or a careful one. Many MP3s record the encoder's name in a tag inside the file, and free tools such as MediaInfo display it.
Bitrate: the size of the budget
Bitrate is the amount of data the encoder may spend on each second of sound. More data lets it describe the music more precisely. The gains shrink as the number rises. The step from 96 to 128 kbps is plain to hear. The step from 256 to 320 kbps is not. The table shows the size per minute and what to expect at five bitrates.
| Bitrate | Size per minute | What to expect |
|---|---|---|
| 96 kbps | 0.72 MB | Fine for speech, thin for music |
| 128 kbps | 0.96 MB | Acceptable, with flaws on good headphones |
| 192 kbps | 1.44 MB | Hard to fault for most listeners |
| 256 kbps | 1.92 MB | Very close to the source |
| 320 kbps | 2.4 MB | The highest MP3 setting |
Sizes are for the audio alone, and cover art adds a little. They assume one bitrate throughout, whereas variable-bitrate files are sized by their average. The MP3 bitrate guide compares the common settings in more detail.
The other figure in a file's properties, the sample rate, rarely needs attention. Nearly all music MP3s use 44.1 or 48 kHz, and both cover the full range of hearing. The two figures are set apart in bitrate vs sample rate.
Why a label is not proof of MP3 audio quality
A bitrate label tells you how big the budget was. It says nothing about what was fed in. Anyone can re-encode a 128 kbps file at 320 kbps. The new file is two and a half times larger and sounds, at most, the same. A 320 kbps option built from online video is the same case, and MP3Juice and 320kbps works through it.
There is a way to check. Encoders working at low bitrates remove the top of the frequency range, and that gap survives any later re-encoding. Open the file in a free audio editor such as Audacity and switch the track to spectrogram view. A file that came from a low-bitrate source shows an empty band above roughly 16 kHz. A true 320 kbps encode of a full-quality recording reaches about 20 kHz.
Playback: the last link
Differences that are clear on wired headphones in a quiet room can vanish elsewhere. A phone speaker reproduces little bass. A car or a bus adds noise that covers fine detail. Volume misleads as well. The louder of two versions tends to sound better, so set both to the same level before you compare. The page on why a downloaded MP3 sounds different from the stream lists the player settings to check.
Bluetooth adds a step of its own. Wireless headphones do not receive the MP3. The phone decodes it and compresses the sound again with a Bluetooth codec such as SBC or AAC before sending it. Over such a link, the gap between a 192 kbps and a 320 kbps file is very hard to hear.
How to judge a file yourself
A few minutes of careful listening tell you more than the label does. Judging a file by ear takes five steps:
- Check the length against the original. A file that ends early was cut off during the download.
- Listen to the hard parts: cymbals, applause, sibilant vocals and the tail of a reverb. Compression faults show there first, as a swishing or gritty texture.
- Compare with the source at the same moment, on the same headphones.
- Ask someone to switch between two versions without telling you which is playing. If you cannot pick out the same one several times running, the difference does not matter to you.
- If two versions sound the same on the equipment you actually use, keep the smaller one.
Choosing a setting that fits the source
For audio taken from online video, a middle setting such as 192 kbps keeps fresh loss low without wasting space, and 320 kbps only makes the file larger. For music ripped from CD or converted from lossless files, 192 to 256 kbps suits most listening, with 320 kbps available if storage is no concern. For anything you may edit or convert again, keep a lossless copy as well, because every further lossy encode costs a little more. The comparison of MP3 vs FLAC shows what a lossless copy keeps. Where an encoder offers a variable bitrate, it can hold the same quality in a smaller file, as CBR vs VBR MP3 explains.
