Talk:Multi-Band Excitation

PC-based chip
found this http://www.dvsinc.com/products/a2020.htm We6jbo 20:59, 3 April 2007 (UTC)

Inaccurate description
From other descriptions I'd say this is currently inaccurate, and that's being charitable. Specifically, one of the advantages of AMBE over other schemes such as CELP is that it does not use codebooks, which are computationaly intensive. There's an article here http://www.eetasia.com/ARTICLES/1999APR/1999APR08_DSP_MSD_TAC2.PDF but it's far too long and I don't know if it is accurate. Chann94501 18:30, 25 October 2007 (UTC)

I would also like to add a reference to this page but as I'm a brand new user, I am hesitant to actually do it because I don't want to mess anything up.

The referennce to add is here:

https://www.researchgate.net/publication/242127600_ROBUST_MULTIBAND_EXCITATION_CODING_OF_SPEECH_BASED_ON_VARIABLE_ANALYSIS_FRAME_SIZES

The article fails to address or link to information that concerns how all editions of multi-band excitation actually works.

In brief summary, it starts by analyzing the signal to be encoded using spectral analysis. The audio spectrum is divided into a certain number of frequency bands (hence the "multi band" in "multi band excitation) and any audio information within that band is measured for its audio level and that value assigned to that band.

This is the general principle, that the audio spectrum is divided into frequency bands having known center frequencies, and the signal level within those bands is measured. The audio signal is reproduced by generating signals in the same bands at the required levels.

By using a fixed known set of band frequencies, and simply measuring the signal strength with them, then the need to encode frequency information into the data stream is eliminated.

A very simple way to see how this works would be to observe an audio spectrum analyzer, as found in many older stereo systems, and observe that it is divided into frequency bands and shows the level of each one with a bar graph display. So the audio signal is represented with a small amount of data. Imagine that the MBE encoder is functioning very much like that 12 band audio spectrum analyzer.

The number of excitation bands and the center frequencies of each band, and the range of frequencies each covers, is important to both voice quality and required data bandwidth. The algorithms in use are under a constant process of refinement with the intention of improving voice quality, robustness of signal, and reduction of bandwith.

Since the output of the voiceband processor now represents audio as a limited number of values in a limited number of frequency bands,  this limited dataset is the symbol bank that is incorporated into the IMBE/AMBE codecs. — Preceding unsigned comment added by Harry Ashcroft (talk • contribs) 16:04, 3 July 2020 (UTC)

Neutrality
The tone of this article speaks too highly of the technology using phrases such as "very powerful." --KJRehberg (talk) 16:19, 6 February 2008 (UTC)
 * That phrase is an example of puffery. Ndanielm (talk) 02:36, 5 November 2011 (UTC)

DMR
If memory serves me, DMR also uses AMBE.

47.187.170.200 (talk) 20:28, 2 February 2018 (UTC)

External links modified (February 2018)
Hello fellow Wikipedians,

I have just modified one external link on Multi-Band Excitation. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20061017004427/http://www.dvsinc.com/products/software.htm to http://www.dvsinc.com/products/software.htm

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 01:34, 8 February 2018 (UTC)