dBpoweramp Speex Codec

Speex is a voice optimized codec, that is it is very efficient at encoding speech, it should not be used for non-voice (ie music) encoding. Speex is Open Source and forms part of the Ogg Vorbis stable. These days Opus has taken over from Speex.

  • Encoding: Yes  [.spx]
  • Multi-processor Encoding: Yes  (with dBpoweramp reference)
  • Decoding: Yes   [.spx]
  • ID Tag Reading: Yes   [Vorbis Comments]
  • ID Tag Writing: Yes   [Vorbis Comments]
  • Unicode Tagging: Yes
  • Supports Album Art: No
  • Gapless Encoding & Decoding: No

Compression Options

Speex offers 3 types of encodings:

  • Quality Target Variable Bit Rate (VBR) VBR files are made up from 100's of small audio chunks, called frames. Whilst encoding VBR, the encoder decides which bit rate to use for each frame. Bit rates can drop down to lower values when it is warranted (if there is not much sound complexity) and switch up to a higher bit rate when required. VBR files should be used when encoding unless there is a specific reason not to,
  • Average Bit Rate (ABR) a little like VBR except constraints are placed on Bit Rate so that over time it averages out to the desired value,
  • Constant Bit Rate (CBR) a constant bit rate is used throughout the encoding process.
Frequency number of samples per second to be encoded, [automatic] allows the output frequency to be set the same as the input frequency (recommended), whilst limiting to maximum and minimum allowed frequencies.

Advanced Options

Voice Activity Detection - if a period of silence is encountered the block is marked as silence instead of wasting bits encoding nothing. VAD is always used for VBR encoding.

File Based Discontinuous Transmission (DTX) - Discontinuous transmission is an addition to VAD/VBR operation, that allows to stop transmitting completely when the background noise is stationary. In file-based operation, since we cannot just stop writing to the file, only 5 bits are used for such frames (corresponding to 250 bps).

Mono Down mix creates a single channel Speex file, if source channels are greater than 1 then they are combined into one.

Complexity set encoding speed/quality trade-off. The higher the value of complexity, the slower the encoding.


Codec Background

Created by Jean Marc Valin, specifically targeted for Voice over IP (VoIP) and audio books.
Further details: http://www.speex.org


Terminology

Encoding: compress and write an audio file,
Decoding: uncompress, or read an audio file,
ID Tags: meta data such as artist & album are embedded inside the audio file,
Lossless: compression without audio quality loss,
Lossy: audio quality is sacrificed (how much depends on bitrate and codec used) to achieve smaller files,
Gapless: allows the decoder to decode audio stream without gaps (silence),
Explorer Audio Popup: a dBpoweramp function, hold the mouse over a supported audio file and details contained are displayed,
Multi-processor Encoding: for dual processor, or multi core processors multiple files can be compressed at once fully using both (or more) CPUs.
Command Line: text interface, where commands are typed (start >> run >> cmd  to get to the command line)

  Main Index