Read seperate .jsonl.gz from fbank filter them and make a Cutset into single variable.

[sanjuktasr]

code :
for items in mf2:
test_item_cuts = librispeech.test_clean_cuts(items)

       # test_filtered_cuts = test_item_cuts.filter(remove_short_and_long_utt)

        if test_item_cuts is not None:
            print(items)
            test_filtered_cuts = test_item_cuts.filter(remove_short_and_long_utt)
            if count==0:
                #test_other_cuts = librispeech.test_other_cuts()
                test_clean_cuts = test_filtered_cuts
            else:
                test_clean_cuts = test_clean_cuts + test_filtered_cuts
            count=count+1
    print(count)

    test_clean_cuts.to_file("manifest.jsonl.gz")

Error:
File "/NAS1/sanjukta_repo_falcon2/zipformer_gitlab_local/zipformer_v1/icefall/egs/librispeech/ASR/./zipformer/decode.py", line 1224, in
main()
File "/usr/local/lib/python3.9/dist-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/NAS1/sanjukta_repo_falcon2/zipformer_gitlab_local/zipformer_v1/icefall/egs/librispeech/ASR/./zipformer/decode.py", line 1162, in main
test_clean_cuts.to_file("manifest.jsonl.gz")
File "/workspace/lhotse/lhotse/serialization.py", line 579, in to_file
store_manifest(self, path)
File "/workspace/lhotse/lhotse/serialization.py", line 564, in store_manifest
manifest.to_jsonl(path)
File "/workspace/lhotse/lhotse/serialization.py", line 346, in to_jsonl
save_to_jsonl(self.to_dicts(), path)
File "/workspace/lhotse/lhotse/serialization.py", line 171, in save_to_jsonl
for item in data:
File "/workspace/lhotse/lhotse/cut/set.py", line 688, in
return (cut.to_dict() for cut in self)
File "/workspace/lhotse/lhotse/cut/data.py", line 77, in to_dict
d["recording"] = self.recording.to_dict()
File "/workspace/lhotse/lhotse/audio/recording.py", line 339, in to_dict
d["transforms"] = [t.to_dict() for t in self.transforms]
File "/workspace/lhotse/lhotse/audio/recording.py", line 339, in
d["transforms"] = [t.to_dict() for t in self.transforms]
AttributeError: 'dict' object has no attribute 'to_dict'

[pzelasko]

Update to the latest lhotse version (ideally from the master branch)

[sanjuktasr]

Can you please elaborate how will updating to the latest version help in achieving the forming a single cutset(<class 'lhotse.lazy.LazyManifestIterator'>) from a list of cutset(<class 'lhotse.lazy.LazyManifestIterator'>)?

[pzelasko]

We had broken transform serialization after some refactoring that I fixed fairly recently (I think 1.24.2 has those fixes already).

d["transforms"] = [t.to_dict() for t in self.transforms]
AttributeError: 'dict' object has no attribute 'to_dict'