What Audio Format Has Most Compression?

Compressed formats are ubiquitous because they make storing and sharing digital audio more efficient. Certain formats compress input audio data more than others. But which audio format has the most compression?

The AMR audio format has the most powerful compression capability. This lossy format is for encoding speec and can compress data at bitrates as low as 4.75kbps. HE-AACv2 and ATRAC3plus have nearly the same compression efficiency as AMR but produce music-level sound quality, even at low bit rates.

Data compression formats enable the storage and sharing of unprecedented amounts of audio information. There are dozens of compression encoders available for professional and home use, some offering significantly more data reduction power than others. Here is the low-down on high-compression audio formats.

AMR: The Audio Format With The Most Compression

The question of which audio format has the most compression is complicated by numerous confounding variables like algorithms, sampling rate, and bit depth. Nevertheless, Adaptive Multi-Rate Audio Codec (AMR) is probably the strongest contender for the audio format with the highest compression capability.

AMR is a lossy compression codec used to encode narrowband speech data. The codec excels in scenarios where sound quality is less of a priority than file size and bandwidth consumption. For instance, AMR is employed widely for storing voice recordings on mobile phones.

One can appreciate AMR’s compression efficiency or power by considering the codec’s typical bitrates for encoding speech data. Bit rates are the conventional proxy measurement for compression efficiency (since the sampling rate and bit depth determine the amount of compression in any given instance).

AMR employs lossy encoding to compress audio data at eight exceedingly low bit rates. In general, AMR compresses input data at between 7.4kbps to 12.2kbps. The format can, however, compress audio data at bitrates as low as 4.75kbps, which is below the 7.4kbps toll quality threshold for speech.

By comparison, most lossy encoders compress audio data at 128kbps to 320kbps (though they can compress at bitrates as low as 32kbps). Lossless encoders compress data between 400kbps and 1411kbps, considerably less compression than AMR.

How Does AMR Work?

AMR uses lossy compression to reduce the size of input audio data. Lossy is a category of compression methods that employ perceptual coding to discard a portion of the inessential source data to decrease the size of the newly encoded file.

AMR samples speech data at a frequency of 160 samples per 20-millisecond frame (8kHz/13-bit). The codec filters the frames to the narrowband speech range of 200Hz to 3400Hz.

AMR Coding

Other lossy formats like MP3 and AAC use transform coding like the modified discrete cosine transform (MDCT) algorithm. AMR, in contrast, employs linear predictive coding (LPC) algorithms similar to lossless encoders like the Free Lossless Audio Codec (FLAC).

As a hybrid encoder, AMR uses LCP to transfer a combination of waveform signal and speech parameter information.

AMR first synthesizes speech from the residual waveform using LPC, encoding the parameters in line spectral pairs (an improved technique to direct quantization). Secondly, AMR codes the residual waveform with the algebraic code-excited linear prediction (ACELP) speech algorithm (a distinct type of LPC).

The AMR format also uses other digital speech processing techniques that decrease the bandwidth a file consumes during moments when audible signals are absent. These techniques are:

comfort noise generation,
discontinuous transmission,
voice activity detection.

While AMR has the most compression compared with other formats, it operates at such low bit rates that this lossy format is only suitable for simple voice data. It might be worth highlighting a few alternative audio formats with high compression power but superior sound quality, allowing their use for more complex music data.

Audio Formats With High Compression Efficiency And Sound Quality

HE-AACv2 and ATRAC 3 are two audio encoding formats that perform powerful lossy compression. Unlike AMR, however, these formats encode and play music because they produce high sound quality at low bit rates.

HE-AACv2

The second version of High-Efficiency Advanced Audio Coding (HE-AAC) has the highest compression efficiency of any audio format that supports music data.

This lossy format has a maximum compression ratio of close to 30:1. Compression ratio is a measure of how efficiently (or powerfully) an encoder compresses audio data and derives from the original data size divided by the size of the compressed data. So, HE-AACv2 creates files nearly 30 times smaller than the original.

HE-AACv2 enhances the compression efficiency and sound quality of its earlier AAC predecessors because it incorporates spectral band replication (SBR) and parametric stereo.

Using SBR, the format produces comparable quality to the core version of AAC at a roughly 50% lower bitrate. HE-AACv2 further extends its compression capabilities by deploying parametric stereo to represent the stereo images of source audio signals.

While HE-AACv2 compresses audio data significantly, the format produces a reasonable level of sound quality at low bitrates (even as low as 24kbps). For this reason, HE-AACv2 is used widely in professional applications with low-bandwidth requirements such as digital audio streaming and broadcasting.

The HE-AAC format is compatible with most commonly-used operating systems and many popular media player programs. Nonetheless, the average user does not require such powerful compression capabilities and is generally more concerned with sound quality.

ATRAC3 And ATRAC3plus

Adaptive Transform Acoustic Coding (ATRAC) is a class of branded (proprietary) audio compression algorithms developed by Sony for its gaming and mobile music devices.

There are multiple versions of ATRAC, with the most advanced ones offering impressive compression efficiency and sound quality. The first ATRAC codec had a compression ratio of 5:1. More recent, state-of-the-art versions of the format, such as ATRAC3 and ATRAC3plus have compression ratios of 10:1 and 20:1 respectively!

ATRAC3 is a hybrid sub-band MDCT lossy compression encoder. When compressing source data, the encoder splits the input signal into four parts (sub-bands) according to their frequency and encodes them using the MDCT algorithm.

ATRAC3 resembles MP3 in its design and functions due to its use of a hybridized sub-band MDCT algorithm and provides comparable compression power. It also lacks the coding efficiency of improved formats like AAC and WMA, which use a pure MDCT algorithm.

ATRAC3plus is a superior version of ATRAC3. This encoder uses sub-band MDCT coding to compress audio data, but also applies Generalized Harmonic Analysis (GHA), a tonal extraction technique that enhances the output sound quality.

AMR, HE-AACv2, And ATRAC3 Vs Other Compression Formats

To appreciate of efficiency of AMR, HE-AAC, and ATRAC3, it is worth considering the bit rates and compression ratios of other audio formats.

MP3 is an efficient lossy compression format but does not compress audio as much as AMR, ATRAC3, and HE-AACv2. The maximum compression ratio of MP3 is 12:1, and the lowest bitrate it encodes at is 32kbps (though 128kbps is generally the minimum for listenable music quality on MP3).

Lossless compression formats perform significantly less compression than lossy encoders. The bit rate for lossless compression codecs ranges from 400kbps to 1411kbps.

Similarly, the average compression ratio for lossless codecs is between 2:1 and 2.5:1. For instance, the Apple Lossless Audio Codec (ALAC) has a maximum compression ratio of 2:1. FLAC, another widely-used lossless codec, also has a maximum compression ratio of 2:1.

The following table summarizes the compression efficiency of these commonly-used audio formats, as defined by their lowest possible bitrates.

Audio Format	Lowest Bitrate
AMR	4.75kbps
HE-AACv2	24kbps
ATRAC3	32kbps
ATRAC3plus	32kbps
MP3	32kbps
FLAC	400kbps
ALAC	400kbps

Conclusion

Determining the audio format with the most compression is complicated when considering the myriad of complex mathematical variables involved in audio data coding. Nonetheless, it is reasonable to conclude that AMR is the audio format with the most compression, encoding speech data at bit rates as low as 4.75kbps!

Audio formats like HE-AACv2 and ATRAC3plus have slightly lower compression efficiency than AMR. However, these formats produce significantly higher sound quality than AMR, making them suitable for compressing music data.

What Audio Format Has Most Compression?