Lossy: Difference between revisions
(Better lead note, sorted formats) |
Artoria2e5 (talk | contribs) |
||
| (13 intermediate revisions by 6 users not shown) | |||
| Line 1: | Line 1: | ||
'''Lossy''' compression is a form of compression that significantly reduce multimedia file size by throwing away information imperceptible to humans. | '''Lossy''' compression is a form of compression that significantly reduce multimedia file size by throwing away information imperceptible to humans. | ||
Human audio perception is not perfect. Lossy compression takes advantage of this | Human audio perception is not perfect. Lossy compression takes advantage of this characteristic. By selective discarding, much unnecessary information is thrown away. The amount of information discarded is usually adjustable, giving a compromise between smaller size with less quality and larger size with higher quality. | ||
The downside to this is that waveform reconstructed from compressed information will never exactly match the original waveform. | The downside to this is that waveform reconstructed from compressed information will never exactly match the original waveform. | ||
==Does Lossy Encoding Preserve Surround Information?== | == Does Lossy Encoding Preserve Surround Information? == | ||
It's better to first ask "does lossy encoding preserve localization" at all. At high bitrates, yes. At lower bitrates, phase information gets sacrificed first, so the stereo image suffers. This applies also to surround-in-stereo formats (Dolby Pro Logic; [https://hydrogenaud.io/index.php/topic,4639 More discussion]), ambisonics-in-stereo (UHJ), and raw B-format ambisonic.<ref>Phase/ambisonic issue discussed in: Mahé, Pierre; Ragot, Stéphane; Marchand, Sylvain (2 September 2019). ''[https://hal.science/hal-02289558 First-Order Ambisonic Coding with PCA Matrixing and Quaternion-Based Interpolation]''. 22nd International Conference on Digital Audio Effects (DAFx-19), Birmingham, UK. p. 284.</ref> | |||
Why does this happen? | |||
* Obviously, the [[bitrate]] controls how much sacrifice happens. Use higher bitrate to prevent this from happening. The lower the bitrate, the worse you can expect the surround imaging become. | |||
* Multi-mono encoding processes each channel separately with no regard to the phase relationship between each channel, so phase errors are more likely to happen. Use [[joint stereo]] (or matrix demixng for more channels). | |||
Mid/Side stereo of [[LAME]] or [[Advanced Audio Coding|AAC]] does not destroy surround information. Also [[Musepack|MPC]] preserves surround information with standard settings reasonably well. | |||
==List of common lossy formats== | == List of common lossy formats == | ||
* [[ | * [[Advanced Audio Coding]] (AAC, also improperly known as [[MP4]] or [[M4A]]) | ||
* [[AC3]] | * [[AC3]] | ||
* [[ATRAC3]] | * [[ATRAC3]] | ||
| Line 18: | Line 21: | ||
* [[MP2]] | * [[MP2]] | ||
* [[MP3]] | * [[MP3]] | ||
* [[Musepack]] (also known as | * [[Musepack]] (also known as MPC, formerly known as MPEGplus or MP+) | ||
* [[Ogg Vorbis]] | * [[Opus]] | ||
* (Ogg) [[Vorbis]] | |||
* [[QDesign]] | * [[QDesign]] | ||
* [[Speex]] (speech only) | * [[Speex]] (speech only) | ||
* [[VQF]] | * [[VQF]] | ||
* [[ | * [[Windows Media Audio]] (WMA) | ||
==See Also== | == See Also == | ||
* [[Choosing the best codec]] | |||
* [[Lossless]] | * [[Lossless]] | ||
[[Category:Codecs|*]] | |||
<references/> | |||
Latest revision as of 02:17, 13 July 2023
Lossy compression is a form of compression that significantly reduce multimedia file size by throwing away information imperceptible to humans.
Human audio perception is not perfect. Lossy compression takes advantage of this characteristic. By selective discarding, much unnecessary information is thrown away. The amount of information discarded is usually adjustable, giving a compromise between smaller size with less quality and larger size with higher quality.
The downside to this is that waveform reconstructed from compressed information will never exactly match the original waveform.
Does Lossy Encoding Preserve Surround Information?
It's better to first ask "does lossy encoding preserve localization" at all. At high bitrates, yes. At lower bitrates, phase information gets sacrificed first, so the stereo image suffers. This applies also to surround-in-stereo formats (Dolby Pro Logic; More discussion), ambisonics-in-stereo (UHJ), and raw B-format ambisonic.[1]
Why does this happen?
- Obviously, the bitrate controls how much sacrifice happens. Use higher bitrate to prevent this from happening. The lower the bitrate, the worse you can expect the surround imaging become.
- Multi-mono encoding processes each channel separately with no regard to the phase relationship between each channel, so phase errors are more likely to happen. Use joint stereo (or matrix demixng for more channels).
Mid/Side stereo of LAME or AAC does not destroy surround information. Also MPC preserves surround information with standard settings reasonably well.
List of common lossy formats
- Advanced Audio Coding (AAC, also improperly known as MP4 or M4A)
- AC3
- ATRAC3
- DTS
- MP2
- MP3
- Musepack (also known as MPC, formerly known as MPEGplus or MP+)
- Opus
- (Ogg) Vorbis
- QDesign
- Speex (speech only)
- VQF
- Windows Media Audio (WMA)
See Also
- ↑ Phase/ambisonic issue discussed in: Mahé, Pierre; Ragot, Stéphane; Marchand, Sylvain (2 September 2019). First-Order Ambisonic Coding with PCA Matrixing and Quaternion-Based Interpolation. 22nd International Conference on Digital Audio Effects (DAFx-19), Birmingham, UK. p. 284.