> The 2500 bit/s mode doesn't use differential (in time of freq) > quantisation, but the 1400 bits/version does. I am yet to be convinced > that bit errors will cause a serious problem - we need to do some tests > I think.
Keep in mind that using fractional bits for codebook entries, etc., will increase your susceptibility to bit errors. If you're using something like a non-binary arithmetic coder, even with a uniform probability model, to pack the symbols, then a single bit error can desync the state of the decoder and cause the rest of the symbols in the packet to be read incorrectly. > LSPs don't directly map to harmonic amplitudes, but loosely to peaks in > the amplitude spectrum. A closely spaced pair of LSPs will map to a peak > in the amplitude envelope. This envelope is like a dotted line tracing Despite that, Rick, you can still weight the codebook search in the LSP domain. There was some discussion on the list about this back in June. See https://sourceforge.net/mailarchive/forum.php?thread_name=4E012793.90600%40perens.com&forum_name=freetel-codec2 > It is possible to synthesise wideband speech(beneath 300Hz, about 4kHz) > without any additional transmitted information based on what we know of > the properties of speech. IIRC, you need roughly a 16th order LPC filter to do a decent job for wideband, and codec2 is currently only using a 10th order filter (which is fine for narrowband). Though at codec2 rates you may be willing to live with less than what others would consider "decent". ------------------------------------------------------------------------------ All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-novd2d _______________________________________________ Freetel-codec2 mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/freetel-codec2
