AudioUtils

Audio Bitrate Explained: What It Means for Quality

Understand audio bitrate, how it affects sound quality and file size, and how to choose the right bitrate for your needs.

Bitrate is the most decision-relevant number in any audio file. It tells you how many bits of data the encoder spends to describe each second of sound, and it is the single biggest lever between "tiny file" and "transparent quality." Pick the wrong bitrate and you either waste storage or ship audible artifacts. Pick the right one and listeners cannot tell the original from the compressed copy.

This guide goes deeper than the usual "higher is better" articles. It covers the math, the encoder behavior, the real thresholds where humans can ABX-distinguish lossy from lossless, and the bitrates the major streaming services actually use.

What Bitrate Actually Measures

Bitrate is the volumetric flow rate of audio data, expressed in kilobits per second (kbps) or megabits per second (Mbps). A 320 kbps MP3 carries 320,000 bits, or 40,000 bytes, for every second of playback. A 64 kbps Opus stream carries 8,000 bytes per second. The unit is bits, not bytes — divide by 8 to get bytes per second, then multiply by duration in seconds to get file size.

Three numbers fully describe uncompressed PCM audio: sample rate, bit depth, and channel count. Multiply them and you get the raw bitrate. CD audio is 44,100 samples per second, 16 bits per sample, 2 channels:

44,100 x 16 x 2 = 1,411,200 bits per second = 1,411 kbps

That 1,411 kbps is the ceiling for standard CD-quality PCM audio and it is also the bitrate of an uncompressed WAV file. Studio-grade 24-bit / 96 kHz stereo runs at 4,608 kbps. A 7.1 channel 24-bit / 48 kHz Blu-ray mix runs at 9,216 kbps. Lossy codecs exist to bring these numbers down by an order of magnitude or more without an audible cost.

Bitrate, Sample Rate, and Bit Depth Are Not the Same Thing

These three terms get used interchangeably, and they should not be. They measure different axes of audio.

  • Sample rate is how often the analog signal is measured per second, in hertz. It sets the maximum frequency you can capture (the Nyquist limit, half the sample rate). For full discussion see the sample rate guide.
  • Bit depth is how precisely each sample is recorded — 16 bits gives 65,536 possible amplitude values, 24 bits gives 16.7 million. Bit depth controls dynamic range and the noise floor.
  • Bitrate is the output of the encoder. For uncompressed PCM, bitrate equals sample rate x bit depth x channels. For compressed audio, bitrate is set independently by the encoder and reflects how aggressively the lossy or lossless algorithm reduced the data.

You can have two MP3 files at the same bitrate (192 kbps) but different sample rates (44.1 kHz vs 22 kHz). You can have a high-bitrate file that sounds worse than a low-bitrate file from a better codec. Bitrate alone does not determine quality — codec, encoder implementation, and source material all matter.

How Lossy Encoders Spend Their Bits

Every lossy codec — MP3, AAC, Opus, Vorbis — uses a psychoacoustic model. The model tracks two phenomena. Frequency masking: a loud tone hides quieter tones at nearby frequencies. Temporal masking: a loud transient hides quieter sounds for roughly 5 ms before and 100-200 ms after. The encoder identifies what is masked, allocates fewer bits to those parts of the spectrum, and quantizes them more coarsely. Whatever bit budget the bitrate provides gets spent on the parts the model believes you will hear.

At low bitrates the budget runs out fast. The encoder collapses stereo to joint stereo or mono, low-passes the signal at 12-15 kHz, and quantizes harshly. You hear the result as warbling cymbals, swirly reverb tails, smeared transients, and a metallic edge on sibilants. At high bitrates the encoder has room to keep stereo separation, hold the full 20 kHz range, and quantize finely. The result becomes transparent — most listeners cannot ABX-distinguish it from the source.

The Transparency Thresholds

Public ABX testing (Hydrogen Audio's listening tests are the gold standard) establishes rough transparency points for each codec on typical music:

  • MP3 (LAME encoder) — transparent for most listeners around 192-256 kbps VBR. Below 128 kbps, artifacts are audible on most material.
  • AAC-LC — transparent around 128-160 kbps. About 30 percent more efficient than MP3.
  • Opus — transparent around 96-128 kbps for stereo music. The current state of the art.
  • Vorbis (aoTuV) — transparent around 128-160 kbps. Roughly equivalent to AAC.
  • HE-AAC v2 — usable down to 32 kbps for stereo with parametric tools, designed for low-bandwidth streaming.

These are listener thresholds for trained ears in quiet environments with good playback equipment. On laptop speakers in a coffee shop, the floor drops considerably. The distinction between 128 kbps and 320 kbps MP3 matters less than people often assume on consumer hardware.

CBR, VBR, ABR — How the Bitrate Is Distributed

Three modes govern how an encoder allocates its bit budget across time.

Constant Bitrate (CBR) locks every frame to the same number of bits. A 192 kbps CBR MP3 uses 192 kbps for the silent intro, the dense climax, and everything between. The advantage is predictable file size and accurate seeking in old hardware decoders. The disadvantage is wasted bits in simple passages and starved bits in complex ones. CBR is still the right choice for live streaming where the network needs a fixed rate.

Variable Bitrate (VBR) lets the encoder allocate more bits to demanding passages and fewer to simple ones, targeting a quality level instead of a fixed rate. LAME's V0 setting averages around 245 kbps but spends 320 kbps on dense rock and 180 kbps on a piano recital. VBR almost always sounds better than CBR at the same average size. The full comparison lives in the VBR vs CBR MP3 explainer.

Average Bitrate (ABR) is the compromise — it varies per frame like VBR but stays close to a target average like CBR. Useful when you need a predictable file size but want some adaptive intelligence.

For new lossy encodes, prefer VBR. For podcast feeds where some legacy apps still mishandle VBR seeking, CBR remains a defensible choice for speech-only content.

Lossless Bitrate Behavior Is Different

FLAC and ALAC do not target a bitrate. They use entropy coding (Rice codes for FLAC) on the difference between predicted and actual samples. The output rate floats with the signal complexity. CD-quality FLAC of a heavy metal track might land at 1,000 kbps. The same FLAC of a sparse vocal might land at 600 kbps. Both are bit-perfect reconstructions of the source.

When you convert MP3 to FLAC, the FLAC bitrate reflects the lossy artifacts in the source — there is nothing to recover. Lossless compression of an already-degraded file just preserves the degradation.

File Size Math, Worked Examples

The formula is simple but worth practicing:

File size (MB) = Bitrate (kbps) x Duration (seconds) / 8,000

A 4-minute song:

  • 64 kbps Opus: 64 x 240 / 8000 = 1.92 MB
  • 128 kbps AAC: 128 x 240 / 8000 = 3.84 MB
  • 256 kbps AAC (Apple Music): 256 x 240 / 8000 = 7.68 MB
  • 320 kbps MP3: 320 x 240 / 8000 = 9.6 MB
  • CD-quality WAV: 1411 x 240 / 8000 = 42.3 MB
  • 24-bit / 96 kHz studio: 4608 x 240 / 8000 = 138 MB

A 60-minute podcast at 64 kbps Opus mono: 28.8 MB. At 128 kbps MP3 stereo: 57.6 MB. The lossy savings dwarf the difference between codec choices.

Bitrate Recommendations by Use Case

The right bitrate depends on what is in the file and where it goes. The bitrate-by-use-case guide breaks each scenario down further; the short version:

  • Voice memos, conference recordings, dictation: 32-64 kbps Opus mono. Speech codecs handle voice down to 16 kbps.
  • Podcasts (speech only): 64-96 kbps Opus or 96-128 kbps MP3 mono. Producers who care about music beds use stereo at 128 kbps MP3.
  • Music for casual listening: 128 kbps AAC or 192 kbps MP3 VBR. Equivalent quality for typical earbuds.
  • Music for serious listening: 256 kbps AAC or 245 kbps VBR MP3 (LAME -V0). Transparent for almost everyone.
  • Archive masters: lossless. FLAC for portability, WAV for DAW work, ALAC for Apple ecosystems.
  • Studio production: 24-bit / 48 kHz or 24-bit / 96 kHz uncompressed PCM during recording and mixing. Mix down to lossy only at delivery.

What the Streaming Services Actually Encode

The bitrates real platforms ship to consumers reveal where the industry has settled.

  • Spotify lists about 24, 96 and 160 kbps in its apps for everyone and about 320 kbps for Premium, plus Lossless up to 24-bit/44.1 kHz FLAC for Premium. Its web player uses AAC at 128 kbps (Free) or 256 kbps (Premium).
  • Apple Music delivers 256 kbps AAC by default, with lossless ALAC options up to 24-bit / 192 kHz on supported tracks at no extra cost.
  • YouTube Music streams 128 kbps AAC or 256 kbps AAC for premium subscribers. The video site itself uses Opus at 50-160 kbps depending on connection.
  • Tidal offers 320 kbps AAC for the standard tier and 16-bit / 44.1 kHz FLAC plus high-resolution MQA for HiFi.
  • Amazon Music HD delivers 16-bit / 44.1 kHz FLAC standard and up to 24-bit / 192 kHz Ultra HD.

The pattern: 128-256 kbps modern lossy is the consumer baseline; lossless is a feature for catalog completeness more than universal everyday use.

Why Higher Is Not Always Better

Three reasons to resist cranking every encode to maximum.

Diminishing returns past transparency. Once a codec hits its transparency threshold, additional bits encode information you cannot hear. A 320 kbps MP3 is 25 percent larger than a 256 kbps MP3, but trained listeners ABX-distinguish them on less than five percent of trials.

Storage and bandwidth costs. A 10,000-track library at 320 kbps MP3 weighs 100 GB. The same library at 192 kbps weighs 60 GB. Distinct only on cold-storage backups, but on a 128 GB phone it is the difference between fitting and not — to shrink an existing library without re-ripping, compress your MP3 files to a lower target bitrate.

Source quality caps everything. If your source is a 96 kbps streaming rip, transcoding to 320 kbps MP3 just embeds the streaming codec's artifacts in a larger container. Bitrate cannot recover information that was never recorded.

Generational Loss When Transcoding

Each lossy-to-lossy transcode applies a fresh psychoacoustic pass. Quantization errors compound. Even 320 kbps MP3 to 256 kbps AAC measurably degrades the signal. The rule: always transcode from the highest-quality source you can find. If you have a FLAC, encode the MP3 from the FLAC, not from another MP3.

Headphones and Speakers Change the Floor

Bitrate decisions interact with playback hardware. Cheap earbuds in a noisy environment mask far more than 128 kbps MP3 ever could — the limiting factor becomes the transducer, not the codec. Studio monitors in a treated room expose every artifact down to bit-perfect lossless. If your audience is mostly Bluetooth earbuds on a commute, 128 kbps AAC is honestly enough. If you are mastering for audiophile streaming, deliver lossless and let the platform downconvert.

Bottom Line

Bitrate determines the quality-versus-size tradeoff in lossy audio and reflects signal complexity in lossless audio. The transparency thresholds are codec-specific: roughly 96 kbps Opus, 128 kbps AAC, 192 kbps MP3 for typical music. CBR is for legacy compatibility; VBR is the default for everything else. Calculate file size with bitrate x duration / 8,000 and you can predict storage to the megabyte. Encode masters lossless and distribute lossy — that single workflow eliminates most bitrate regret.

Frequently Asked Questions

What does kbps mean in audio?

kbps stands for kilobits per second and measures how many thousands of bits the audio file uses to describe each second of sound. A 192 kbps MP3 carries 192,000 bits — roughly 24,000 bytes — every second. Higher kbps means more data per second, which usually means better quality and a larger file. The unit is bits, not bytes; divide by 8 to convert to kilobytes per second. CD-quality uncompressed audio is 1,411 kbps, while modern lossy codecs reach perceptual transparency between 96 and 256 kbps depending on the codec.

What is a good bitrate for MP3?

For music, 192 kbps VBR (LAME setting -V2) is the practical floor for transparent quality on typical listening gear, and 245 kbps VBR (LAME -V0) is the safe choice when storage is not a concern. CBR 320 kbps is the maximum MP3 supports and is overkill for nearly all listeners. For podcasts and speech, 96-128 kbps mono CBR is plenty — voice content has less spectral complexity than music. Below 128 kbps stereo, MP3 audibly compresses cymbals, reverb tails, and high frequencies on most material.

Is 320 kbps better than 256 kbps?

Technically yes, audibly rarely. In blind tests most listeners cannot reliably tell 256 from 320 kbps MP3, and the 320 file is 25 percent larger. AAC at 256 kbps generally sounds equal or better to MP3 at 320 kbps because AAC is the more efficient codec. Pick 320 kbps when you want the maximum MP3 supports and storage is irrelevant; pick 256 kbps AAC if you want better quality at a smaller size; pick lossless if you want bit-perfect reproduction.

How do I calculate audio file size from bitrate?

Multiply the bitrate in kbps by the duration in seconds, then divide by 8,000 to get megabytes. A 4-minute (240-second) track at 192 kbps is 192 x 240 / 8,000 = 5.76 MB. A 60-minute podcast at 128 kbps mono is 128 x 3,600 / 8,000 = 57.6 MB. The formula assumes a clean stream with no container overhead; in practice, MP3 and M4A add 1-3 percent for headers and tags. For variable bitrate files, the average bitrate gives an estimate within a few percent.

Should I use CBR or VBR for MP3?

VBR is the better default for music. It allocates more bits to complex passages and fewer to simple ones, producing better quality at a smaller average size. The encoder targets a quality setting rather than a fixed rate, so the file adapts to the content. CBR remains useful in three cases: live streaming where the network requires a fixed bitrate, podcast feeds where some legacy apps still mishandle VBR seeking, and broadcast workflows that mandate fixed-rate output. For everything else — Spotify uploads, music distribution, archival lossy copies — VBR is preferred.

What bitrate does Spotify use?

Spotify's support page lists about 24, 96 and 160 kbps in the apps for everyone, about 320 kbps (Very high) for Premium, and a Premium Lossless option up to 24-bit/44.1 kHz FLAC. The web player uses AAC at 128 kbps (Free) or 256 kbps (Premium).

Does higher bitrate always mean better quality?

Only up to the codec's transparency threshold and only if the source has the quality to begin with. Past the threshold (around 96 kbps Opus, 128 kbps AAC, 192 kbps MP3), additional bits encode information listeners cannot detect. Bitrate also cannot recover quality that was never captured — encoding a 96 kbps streaming rip to 320 kbps MP3 just inflates the file size while preserving the original codec's artifacts. Codec choice, encoder implementation, source quality, and playback hardware all interact with bitrate to determine perceived quality.