FreeCalypso > hg > freecalypso-docs

annotate DSP-speech-decoder @ 108:e26623146358 default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
new article DSP-speech-decoder
author Mychaela Falconia <falcon@freecalypso.org>
date Tue, 29 Oct 2024 22:11:41 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
108
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
1 As described in Calypso-digital-voice article, we can use Calypso MCSI (an
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
2 auxiliary hardware interface brought out on FreeCalypso dev boards) in the so-
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
3 called "Bluetooth headset" mode to capture the voice call downlink in digital
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
4 form, specifically 16-bit linear PCM, after speech decoding performed by the
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
5 DSP, but before D-to-A conversion for analog audio output. As of 2024-10, we
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
6 have a set of tools (Lattice Icestick FPGA board, a gateware design for this
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
7 FPGA, and corresponding host software tools - see fc-pcm-if Hg repository) that
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
8 puts these ideas into practice; using these tools, we can now answer some long-
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
9 standing questions about exactly how Calypso DSP implements speech decoding for
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
10 various GSM codecs.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
11
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
12 FRv1 decoder implementation
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
13 ===========================
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
14
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
15 The original set of Full Rate codec specs did not include the feature of in-band
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
16 encoder and decoder homing frames; this feature was included in HRv1 and EFR
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
17 from day 1, and was also back-ported to FRv1 in later versions of 06.10 spec -
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
18 but it was not present originally, and it is defined as optional in the later
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
19 spec versions. Going into these Calypso MCSI experiments, we had this question:
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
20 does Calypso DSP implement decoder homing frames for FRv1, or does it not?
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
21 Experimental observation: in-band decoder homing is NOT implemented for FRv1 in
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
22 this DSP. If we feed the DHF defined in later versions of 06.10 spec to our
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
23 Calypso MS via TCH DL, it gets treated as a regular speech frame without any
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
24 special handling; more specifically, we don't see PCM16 output frames consisting
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
25 of 0x0008 words as required by the spec when the optional in-band decoder homing
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
26 feature is present.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
27
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
28 This lack of in-band homing feature makes it difficult to test the decoder
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
29 against official ETSI test sequences, as we have no way of getting it into a
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
30 defined state. The ancient GSM 11.10 test spec provides a mechanism called DAI
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
31 that includes the necessary reset signal, and it appears that Calypso DSP does
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
32 support operating MCSI as DAI rather than "Bluetooth" mode - but we haven't done
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
33 any experiments with this DAI mode yet. Hence our exploration of FRv1 decoder
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
34 implementation in our dear Calypso DSP ends here for now.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
35
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
36 EFR decoder implementation
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
37 ==========================
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
38
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
39 Unlike the situation with FRv1, the in-band codec homing feature is mandatory
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
40 for EFR per the specs. And with Calypso DSP we got good news: it does implement
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
41 this feature at least in the decoding direction, which is the only one we've
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
42 tested so far. Exactly per the spec, the first DHF arriving from TCH DL is
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
43 decoded like a regular speech frame, but the decoder state is reset at the end.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
44 All subsequent DHFs turn into 0x0008 output, and we indeed see the latter coming
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
45 out on MCSI.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
46
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
47 Once we established that the decoder homing feature works as specified, our
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
48 next question was: does our dear Calypso DSP implement the original bit-exact
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
49 definition of EFR, or does it implement what we call the AMR-EFR hybrid?
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
50 Experimental observation: at least in the decoding direction, it implements the
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
51 original bit-exact EFR decoder! This finding is remarkable because all of our
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
52 experiments in this series were performed on DSP ROM version 3606, the final
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
53 one that not only includes AMR support, but has been subject to whatever rework
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
54 TI did between 3416 and 3606 ROMs. If we assume (based on some experiments in
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
55 past years where we disabled loading of TI's official DSP patches) that AMR
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
56 support is already present in the DSP ROM, as opposed to being contained
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
57 entirely in the patches, it follows that the DSP ROM must contain both EFR and
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
58 AMR codec implementations, yet the EFR implementation has not been AMR-ized.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
59 (In contrast, the network-side speech transcoder implementation used by T-Mobile
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
60 USA and Telcel Mexico appears to use maximally-shared code between EFR and AMR,
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
61 resulting in the AMR-EFR hybrid that differs from the original bit-exact
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
62 definition, but has been blessed by 3GPP as an acceptable alternative.)
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
63
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
64 Strange corruption bug
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
65 ----------------------
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
66
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
67 As we fed ETSI test sequences from GSM 06.54 to our Calypso DSP under test via
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
68 TCH DL, checking for an exact match, we observed what appears to be a corruption
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
69 bug of some kind (corrupting either codec parameter bits or state variables)
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
70 that manifests in a manner that seems random. In our test, we concatenated all
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
71 21 test sequences from GSM 06.54 (test0.cod through test20.cod) and fed the
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
72 resulting super-long sequence into TCH DL. The test was performed twice; on
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
73 the second try we had 'tch record' running in fc-shell, capturing TCH DL as
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
74 received by the Calypso MS, and we confirmed that the entire super-long test
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
75 sequence was received without errors. The decoded output (captured via MCSI)
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
76 was then split into 21 separate robe files following frame counts: because each
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
77 test*.cod input sequence begins with two DHFs, the decoder resets between these
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
78 concatenated test sequences, allowing the output to be compared against
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
79 test*.out reference files with per-sequence granularity. The final diff
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
80 comparison between split-out robe files and the official reference decoder
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
81 output revealed the following oddities:
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
82
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
83 * In the first run, decoded outputs for test0, test6, test11, test14, test15,
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
84 test16 and test20 showed a mismatch. The difference in test0 outputs was
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
85 examined closer: the diff begins on a subframe boundary and continues until
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
86 the next DHF. All other test sequences got a perfect match to ETSI reference
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
87 decoder output.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
88
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
89 * In the second run, a different set of sequences showed a mismatch: this time
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
90 the failing sequences were test1, test5, test6, test7, test15 and test16.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
91 The other sequences that produced mismatches in the first run produced
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
92 perfectly good output (matching ETSI reference decoder) in this run!
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
93
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
94 * test6, test15 and test16 were the three sequences that failed in both runs.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
95 However, diffing between the two runs shows that the errors in each of these
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
96 3 twice-failed sequences are different between runs, further strengthening
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
97 the appearance of corruption that seems totally random.
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
98
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
99 We do not currently have any better explanation for this oddity than a data
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
100 corruption bug somewhere in the DSP. For the record, these tests were performed
e26623146358 new article DSP-speech-decoder
Mychaela Falconia <falcon@freecalypso.org>
parents:
diff changeset
101 on an FCDEV3B with Calypso DSP ROM version 3606 and DSP patch version 6840.