Modify asr.sh

[kamo-naoyuki]

This PR includes my self-answer to #4944.

• I changed asr.sh to raise an error if --train_set equal to --valid_set, or --train_set is also included in --test_sets.
• If --valid_set is included in --test_sets, --eval_valid_set option is enabled.
• if --eval_valid_set is enabled, dump/org/${valid_set} is evaluated at the decoding stage instead of dump/${valid_set}

Someone might still think we should filter the long/short utterances in the training python process, but finally, I concluded it's a bad idea. Please let me go in this direction.

I also changed the behavior of the --skip_train option:

• Before this PR: The data preparation for ${train_set} and ${valid_set} still works. This is inconvinient if using a pre-trained model and evaluation is only required.
• After this PR: The data preparation for ${train_set} and ${valid_set} can be also skipped.

Finally, I added a new feat-type: --feats_type raw_copy. This is almost the same as --feats_type raw_copy, but it can skip format_wav_scp.py.

This type is useful if the file format is already correct.　Sometimes, we create a new evaluation set from an existing dataset, e.g. applying a speech enhancement method. In this case. the data set may be already correct, so we might want to skip
format_wav_scp.py.

Please note that if an user specifies --feats_type raw_copy, the user is responsible to guaranteed that the data format follows correctly our requirements.

[codecov]

Codecov Report

Merging #5020 (09ce640) into master (7964a2a) will increase coverage by 3.55%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #5020      +/-   ##
==========================================
+ Coverage   73.47%   77.02%   +3.55%     
==========================================
  Files         606      606              
  Lines       53748    53748              
==========================================
+ Hits        39491    41400    +1909     
+ Misses      14257    12348    -1909     

Flag	Coverage Δ
test_integration_espnet1	66.29% <ø> (+<0.01%)	⬆️
test_integration_espnet2	47.96% <ø> (?)
test_python	66.85% <ø> (+0.02%)	⬆️
test_utils	23.28% <ø> (ø)

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

see 61 files with indirect coverage changes

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

[kamo-naoyuki]

@popcornell

Now we can use the same valid_set and test_set without modification. If ${valid_set} is included in ${test_sets}, it is just replaced to org/${valid} in the evaluation stages.

As the above comment, I also modified the behaviour of --skip_train. If --skip_train true is specified, the data preparation for the train_set can be skipped. This is useful for the chime7 challenge.

[popcornell]

This is great !