Fix segmenting for pcla inference (NVIDIA#5849)

* Fix segmenting for pcla inference

Signed-off-by: Matvei Novikov <mattyson.so@gmail.com>

* Fix segmenting for pcla inference

Signed-off-by: Matvei Novikov <mattyson.so@gmail.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

Signed-off-by: Matvei Novikov <mattyson.so@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

examples/nlp/token_classification/punctuate_capitalize_infer.py

@@ -36,7 +36,7 @@
     --input_manifest <PATH/TO/INPUT/MANIFEST> \
     --output_manifest <PATH/TO/OUTPUT/MANIFEST> \
     --use_audio
-    
+
 
 <PATH/TO/INPUT/MANIFEST> is a path to NeMo ASR manifest. Usually it is an output of
     NeMo/examples/asr/transcribe_speech.py but can be a manifest with 'text' key. Alternatively you can use
@@ -55,7 +55,7 @@ def get_args() -> argparse.Namespace:
         formatter_class=argparse.ArgumentDefaultsHelpFormatter,
         description="The script is for restoring punctuation and capitalization in text or text and audio. To use text and audio use '--use_audio'. Long strings are split into "
         "segments of length `--max_seq_length`. `--max_seq_length` is the length which includes [CLS] and [SEP] "
-        "tokens. Long audios are split into segments of length 4000*`--max_seq_length`. Parameter `--step` controls segments overlapping. `--step` is a distance between beginnings of "
+        "tokens. If `--use_audio` is set, samples with texts longer than `--max_seq_length` will be ignored. Parameter `--step` controls segments overlapping. `--step` is a distance between beginnings of "
         "consequent segments. Model outputs for tokens near the borders of tensors are less accurate and can be "
         "discarded before final predictions computation. Parameter `--margin` is number of discarded outputs near "
         "segments borders. Probabilities of tokens in overlapping parts of segments multiplied before selecting the "

nemo/collections/nlp/data/token_classification/punctuation_capitalization_infer_dataset.py

@@ -132,6 +132,9 @@ def get_features_infer(
     for q_i, (query_st, query_audio) in enumerate(zip(st, audios)):
         q_inp_ids, q_segment_ids, q_subtokens_mask, q_inp_mask, q_quantities_of_preceding_words = [], [], [], [], []
         q_audio_queries, q_audio_lengths = [], []
+        if query_audio and length < len(query_st):
+            logging.info(f'Ignoring query with id {q_i}')
+            continue
         for i in range(0, max(len(query_st), length) - length + step, step):
             subtokens = [tokenizer.cls_token] + query_st[i : i + length] + [tokenizer.sep_token]
             q_inp_ids.append(tokenizer.tokens_to_ids(subtokens))
@@ -140,7 +143,7 @@ def get_features_infer(
             q_inp_mask.append([True] * len(subtokens))
             q_quantities_of_preceding_words.append(np.count_nonzero(stm[q_i][:i]))
             if query_audio:
-                samples = query_audio.samples[i * 4000 : (i + length) * 4000]
+                samples = query_audio.samples
                 q_audio_queries.append(samples)
                 q_audio_lengths.append(len(samples))
         all_input_ids.append(q_inp_ids)