FreeCalypso > hg > gsm-net-reveng

annotate doc/TFO-xform/EFR @ 36:d9553c7ac6ea

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
doc/TFO-xform/EFR: beginning of article
author Mychaela Falconia <falcon@freecalypso.org>
date Tue, 03 Sep 2024 07:08:24 +0000
parents
children 4ab7cc414ed2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
36
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
1 TFO transform for EFR
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
2 =====================
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
3
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
4 Unlike the situation with FRv1 and HRv1, the standard endpoint decoder for EFR
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
5 provides no help for implementing a TFO transform. The reference EFR decoder
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
6 source from ETSI includes bad frame handling and Rx DTX functions, but the logic
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
7 that implements these functions is interwoven throughout the body of the decoder
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
8 and does not form a separable front-end. Most saliently, this Rx DTX and ECU
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
9 logic in the reference decoder does not operate on coded parameters as would be
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
10 needed for a TFO transform, instead it operates on linear values deeper in the
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
11 decoder after parameter dequantization.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
12
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
13 Given that Abis is a de facto proprietary interface that is not interoperable
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
14 between different vendors (and the same holds for Ater in those BSS designs
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
15 that separate the TRAU from the BSC), and given how daunting it seems to
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
16 implement a true TFO transform for EFR, prior to getting our Nokia TCSM2 lab
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
17 setup I was wondering if historical TRAU vendors really did implement this
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
18 TFO transform, or if perhaps they used some kind of "cheating" trick on their
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
19 Abis similar to what we did in OsmoBTS in mid-2023. However, once I got our
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
20 Nokia TCSM2 gear working, set up a TFO connection between two active TRAU
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
21 channels in EFR mode and passed some test sequences through it, it became clear
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
22 that Nokia did implement a real "honest-to-god" TFO transform for EFR: the
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
23 TRAU-DL frame stream is 100% valid "speech" frames (no idle frames or other
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
24 aberrations inserted) even when the TRAU-UL stream fed via TFO contains BFI
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
25 speech frames and DTXu pauses - the TRAU really does apply bad frame handling
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
26 and comfort noise insertion on parameter level.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
27
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
28 Seeing that at least one major historical vendor did implement TFO transform
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
29 for EFR, and seeing the output from that transform, has set up a sportive
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
30 challenge for me: I no longer have a valid excuse to not do it. I now have a
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
31 desire to produce a FOSS implementation of TFO transform for EFR in Themyscira
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
32 libraries (probably in libgsmefr), and make it no worse than Nokia's
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
33 implementation in TCSM2.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
34
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
35 Bad frame handling in speech mode
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
36 =================================
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
37
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
38 Looking at the DL speech frames that were synthesized by the TRAU in those
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
39 frame positions where the incoming UL stream via TFO had BFIs, we can make the
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
40 following observations:
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
41
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
42 * The 5 LPC parameters are different in each generated substitution/muting
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
43 frame, hence it looks like the TFO transform is running the quantization
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
44 algorithm for each output frame to produce LPC parameters that aim for the
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
45 substitution/muting LSFs of the official "example solution".
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
46
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
47 * LTP lag parameters remain constant for each run of BFIs between good speech
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
48 frames; the lag value encoded therein matches the LTP lag (integer part only)
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
49 from the 4th subframe of the last good speech frame, just like in the official
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
50 endpoint decoder.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
51
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
52 * Surprising bit: the 4 LTP gain values from the last good speech frame are
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
53 endlessly regurgitated verbatim in each substitution/muting frame, without
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
54 any signs of the attenuation I expected to see based on the official "example
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
55 solution".
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
56
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
57 * Another surprising bit: the 35-bit fixed codebook sequence in each subframe
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
58 is taken from the corresponding subframe of the last good speech frame,
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
59 contrary to the official "example solution" that takes these bits from the
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
60 errored frames.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
61
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
62 * The four fixed codebook gain parameters in the emitted substitution/muting
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
63 frames differ from one frame to the next in the case of multiple BFI frames
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
64 in a row, and they also differ between subframes in the same frame - hence
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
65 these parameters are clearly being regenerated as output progresses. However,
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
66 the quantization algorithm for this parameter is so complex that I haven't
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
67 been able to make a more intelligent analysis yet.
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
68
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
69 Looking at the first good speech frame that follows each BFI substitution/muting
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
70 insert, we see that it is mostly unaltered: no alterations were seen to LPC or
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
71 LTP parameters, in particular. However, in the case of the fixed codebook gain
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
72 parameter we see a different behavioral pattern: most of the time it is also
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
73 unaltered, but sometimes we see reduction in this parameter, and even then it
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
74 is only in certain subframes. Are we perhaps seeing a capping of the fixed
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
75 codebook gain in the first good frame following BFI, similar to that implemented
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
76 in the reference endpoint decoder? A better understanding of the quantization
d9553c7ac6ea doc/TFO-xform/EFR: beginning of article
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
77 mechanism for this parameter will be needed.