Fixing some issues with chime7-task1 baseline

[popcornell]

many thanks to Kamo Naoyuki @kamo-naoyuki .
I think there was a mistake on the close talk microphones for dipco dev (they were removed for eval).
I will also address your other useful comments but I am unsure about the sox one.

[codecov]

Codecov Report

Merging #4925 (ecf3bd3) into master (7cf99cf) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #4925   +/-   ##
=======================================
  Coverage   76.56%   76.56%           
=======================================
  Files         603      603           
  Lines       53753    53753           
=======================================
  Hits        41157    41157           
  Misses      12596    12596           

Flag	Coverage Δ
test_integration_espnet1	66.33% <ø> (ø)
test_integration_espnet2	47.58% <ø> (ø)
test_python	66.44% <ø> (ø)
test_utils	23.35% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[kamo-naoyuki]

@popcornell About dipco, this fix works for me. Thank you!

[kamo-naoyuki]

I got the following error at stage4

Stage 4: Enhance segments using GSS
local/run_gss.sh: line 54: affix[@]: unbound variable

I guess it can be fixed by

"${affix[@]}" -> ${affix[@]:-}

However, I'm not sure why using an array variable i.e. affix=().

[popcornell]

I am using that variable because I want that when channels is undefined it is not passed so GSS uses all available ones.

It runs fine on my side so this is strange. What does ${affix[@]:-} mean ? I am not a great bash expert

[kamo-naoyuki]

I am using that variable because I want that when channels is undefined it is not passed so GSS uses all available ones.

The following is enough for your case. It is not necessary to use array variable.

  affix=
  if ! [ $channels == all  ]; then
    affix+="--channels=$channels"
  fi

It runs fine on my side so this is strange. What does ${affix[@]:-} mean ? I am not a great bash expert

This script obviously can't work with "set -u" if channels != all i.e. $channels is empty.

Please check.

• ${foo:-} means returning a empty string if foo is undefined (This is a workaround in the case of using an empty array variable with set -u.)
• If you give "${foo}" to gss command, it is interpreted as an argument even if foo is empty and it causes an arguments error for gss command. ${foo} can avoid this problem.

[popcornell]

I am starting to add this pretrained model https://huggingface.co/popcornell/chime7_task1_asr1_baseline
WER (%) on (GSS on dev) of

1. CHiME-6 35.5
2. DiPCo 39.3
3. Mixer 6 20.6

Note that now i discarded sentences < 0.2 seconds (not many on dev, so it won't change much) so it is not true WER for the challenge.

[popcornell]

if decode only i run data prep and GSS only on development sets

[YoshikiMas]

I faced minor issues in install_dependencies.sh.
I'm not sure but I think this line could be ${MAIN_ROOT}/tools/installers/install_s3prl.sh.
Also, this line should be ${MAIN_ROOT}/tools/installers/install_gss.sh.

[popcornell]

Indeed thanks Yoshiki !

[sw005320]

LGTM
Once the CI check is passed, I'll merge this.

[popcornell]

Note that I will need to change the README.md too probably based on what we decide tomorrow about the channel selection stuff.

[popcornell]

Rules are not clear not if

espnet/egs2/chime7_task1/asr1/run.sh

Line 129 in ecf3bd3

if [ ${dset_name} == dipco ]; then

should be allowed or not. IMO it should be allowed but then we have to distinguish it from automatic or manual domain classification.
The model remains the same (we want to discourage one model for each scenario) but you just select a bunch of microphones with some a priori knowledge for example of the topology of the array (e.g. linear in chime6 and i use only outer mics)

[sw005320]

OK, please make additional PRs.