add chunk width randomization

egs/libricss/SURT/dprnn_pruned_transducer_stateless7/decode_libricss.py

@@ -518,7 +518,11 @@ def save_results(
         )
         with open(errs_filename, "w") as f:
             wer = write_surt_error_stats(
-                f, f"{test_set_name}-{key}", results, enable_log=True
+                f,
+                f"{test_set_name}-{key}",
+                results,
+                enable_log=True,
+                num_channels=params.num_channels,
             )
             test_set_wers[key] = wer
 

egs/libricss/SURT/dprnn_pruned_transducer_stateless7/dprnn.py

@@ -1,3 +1,4 @@
+import random
 from typing import Optional, Tuple
 
 import torch
@@ -190,6 +191,7 @@ class DPRNN(nn.Module):
         dropout=0.1,
         num_blocks=1,
         segment_size=50,
+        chunk_width_randomization=False,
     ):
         super().__init__()
 
@@ -198,6 +200,7 @@ class DPRNN(nn.Module):
         self.hidden_size = hidden_size
 
         self.segment_size = segment_size
+        self.chunk_width_randomization = chunk_width_randomization
 
         self.input_embed = nn.Sequential(
             ScaledLinear(feature_dim, input_size),
@@ -243,7 +246,11 @@ class DPRNN(nn.Module):
         input = self.input_embed(input)
         B, T, D = input.shape
 
-        input, rest = split_feature(input.transpose(1, 2), self.segment_size)
+        if self.chunk_width_randomization and self.training:
+            segment_size = random.randint(self.segment_size // 2, self.segment_size)
+        else:
+            segment_size = self.segment_size
+        input, rest = split_feature(input.transpose(1, 2), segment_size)
         # input shape: batch, N, dim1, dim2
         # apply RNN on dim1 first and then dim2
         # output shape: B, output_size, dim1, dim2
@@ -291,6 +298,7 @@ if __name__ == "__main__":
         dropout=0.1,
         num_blocks=3,
         segment_size=20,
+        chunk_width_randomization=True,
     )
     input = torch.randn(2, 1002, 80)
     print(model(input).shape)

egs/libricss/SURT/dprnn_pruned_transducer_stateless7/train.py

@@ -104,24 +104,31 @@ def add_model_arguments(parser: argparse.ArgumentParser):
         "--num-mask-encoder-layers",
         type=int,
         default=4,
-        help="Number of layers in the SkiM based mask encoder.",
+        help="Number of layers in the DPRNN based mask encoder.",
     )
 
     parser.add_argument(
         "--mask-encoder-dim",
         type=int,
         default=256,
-        help="Hidden dimension of the LSTM blocks in SkiM.",
+        help="Hidden dimension of the LSTM blocks in DPRNN.",
     )
 
     parser.add_argument(
         "--mask-encoder-segment-size",
         type=int,
         default=32,
-        help="Segment size of the SegLSTM in SkiM. Ideally, this should be equal to the "
+        help="Segment size of the SegLSTM in DPRNN. Ideally, this should be equal to the "
         "decode-chunk-length of the zipformer encoder.",
     )
 
+    parser.add_argument(
+        "--chunk-width-randomization",
+        type=bool,
+        default=False,
+        help="Whether to randomize the chunk width in DPRNN.",
+    )
+
     # Zipformer config is based on:
     # https://github.com/k2-fsa/icefall/pull/745#issuecomment-1405282740
     parser.add_argument(
@@ -508,6 +515,7 @@ def get_mask_encoder_model(params: AttributeDict) -> nn.Module:
         output_size=params.feature_dim * params.num_channels,
         segment_size=params.mask_encoder_segment_size,
         num_blocks=params.num_mask_encoder_layers,
+        chunk_width_randomization=params.chunk_width_randomization,
     )
     return mask_encoder
 

icefall/utils.py

@@ -827,6 +827,7 @@ def write_surt_error_stats(
     test_set_name: str,
     results: List[Tuple[str, str]],
     enable_log: bool = True,
+    num_channels: int = 2,
 ) -> float:
     """Write statistics based on predicted results and reference transcripts for SURT
     multi-talker ASR systems. The difference between this and the `write_error_stats`
@@ -867,20 +868,137 @@ def write_surt_error_stats(
     """
     from meeteval.wer import wer
 
-    wers = []
+    subs: Dict[Tuple[str, str], int] = defaultdict(int)
-    assignments = []
+    ins: Dict[str, int] = defaultdict(int)
-    for cut_id, ref, hyp in results:
+    dels: Dict[str, int] = defaultdict(int)
-        orc_wer = wer.orc_word_error_rate(ref, hyp)
+    ref_lens: List[int] = []
-        wers.append(orc_wer)
-        assignments.append(orc_wer.assignment)
 
-    orc_wer = wer.combine_error_rates(*wers)
+    print(
-    tot_errs = orc_wer.errors
+        "Search below for sections starting with PER-UTT DETAILS:, "
-    ref_len = orc_wer.length
+        "SUBSTITUTIONS:, DELETIONS:, INSERTIONS:, PER-WORD STATS:",
-    tot_err_rate = orc_wer.error_rate
+        file=f,
+    )
+
+    print("", file=f)
+    print("PER-UTT DETAILS: corr or (ref->hyp)  ", file=f)
+
+    # `words` stores counts per word, as follows:
+    #   corr, ref_sub, hyp_sub, ins, dels
+    words: Dict[str, List[int]] = defaultdict(lambda: [0, 0, 0, 0, 0])
+    num_corr = 0
+    ERR = "*"
+    for cut_id, ref, hyp in results:
+        # First compute the optimal assignment of references to output channels
+        orc_wer = wer.orc_word_error_rate(ref, hyp)
+        assignment = orc_wer.assignment
+        refs = [[] for _ in range(num_channels)]
+        # Assign references to channels
+        for i, ref_text in zip(assignment, ref):
+            refs[i] += ref_text.split()
+        hyps = [hyp_text.split() for hyp_text in hyp]
+        # Now compute the WER for each channel
+        for ref_c, hyp_c in zip(refs, hyps):
+            ref_lens.append(len(ref_c))
+            ali = kaldialign.align(ref_c, hyp_c, ERR)
+            for ref_word, hyp_word in ali:
+                if ref_word == ERR:
+                    ins[hyp_word] += 1
+                    words[hyp_word][3] += 1
+                elif hyp_word == ERR:
+                    dels[ref_word] += 1
+                    words[ref_word][4] += 1
+                elif hyp_word != ref_word:
+                    subs[(ref_word, hyp_word)] += 1
+                    words[ref_word][1] += 1
+                    words[hyp_word][2] += 1
+                else:
+                    words[ref_word][0] += 1
+                    num_corr += 1
+            combine_successive_errors = True
+            if combine_successive_errors:
+                ali = [[[x], [y]] for x, y in ali]
+                for i in range(len(ali) - 1):
+                    if ali[i][0] != ali[i][1] and ali[i + 1][0] != ali[i + 1][1]:
+                        ali[i + 1][0] = ali[i][0] + ali[i + 1][0]
+                        ali[i + 1][1] = ali[i][1] + ali[i + 1][1]
+                        ali[i] = [[], []]
+                ali = [
+                    [
+                        list(filter(lambda a: a != ERR, x)),
+                        list(filter(lambda a: a != ERR, y)),
+                    ]
+                    for x, y in ali
+                ]
+                ali = list(filter(lambda x: x != [[], []], ali))
+                ali = [
+                    [
+                        ERR if x == [] else " ".join(x),
+                        ERR if y == [] else " ".join(y),
+                    ]
+                    for x, y in ali
+                ]
+
+            print(
+                f"{cut_id}:\t"
+                + " ".join(
+                    (
+                        ref_word
+                        if ref_word == hyp_word
+                        else f"({ref_word}->{hyp_word})"
+                        for ref_word, hyp_word in ali
+                    )
+                ),
+                file=f,
+            )
+    ref_len = sum(ref_lens)
+    sub_errs = sum(subs.values())
+    ins_errs = sum(ins.values())
+    del_errs = sum(dels.values())
+    tot_errs = sub_errs + ins_errs + del_errs
+    tot_err_rate = "%.2f" % (100.0 * tot_errs / ref_len)
 
     if enable_log:
-        logging.info(f"[{test_set_name}] %WER {tot_errs / ref_len:.2%}")
+        logging.info(
+            f"[{test_set_name}] %WER {tot_errs / ref_len:.2%} "
+            f"[{tot_errs} / {ref_len}, {ins_errs} ins, "
+            f"{del_errs} del, {sub_errs} sub ]"
+        )
+
+    print(f"%WER = {tot_err_rate}", file=f)
+    print(
+        f"Errors: {ins_errs} insertions, {del_errs} deletions, "
+        f"{sub_errs} substitutions, over {ref_len} reference "
+        f"words ({num_corr} correct)",
+        file=f,
+    )
+
+    print("", file=f)
+    print("SUBSTITUTIONS: count ref -> hyp", file=f)
+
+    for count, (ref, hyp) in sorted([(v, k) for k, v in subs.items()], reverse=True):
+        print(f"{count}   {ref} -> {hyp}", file=f)
+
+    print("", file=f)
+    print("DELETIONS: count ref", file=f)
+    for count, ref in sorted([(v, k) for k, v in dels.items()], reverse=True):
+        print(f"{count}   {ref}", file=f)
+
+    print("", file=f)
+    print("INSERTIONS: count hyp", file=f)
+    for count, hyp in sorted([(v, k) for k, v in ins.items()], reverse=True):
+        print(f"{count}   {hyp}", file=f)
+
+    print("", file=f)
+    print("PER-WORD STATS: word  corr tot_errs count_in_ref count_in_hyp", file=f)
+    for _, word, counts in sorted(
+        [(sum(v[1:]), k, v) for k, v in words.items()], reverse=True
+    ):
+        (corr, ref_sub, hyp_sub, ins, dels) = counts
+        tot_errs = ref_sub + hyp_sub + ins + dels
+        ref_count = corr + ref_sub + dels
+        hyp_count = corr + hyp_sub + ins
+
+        print(f"{word}   {corr} {tot_errs} {ref_count} {hyp_count}", file=f)
 
     print(f"%WER = {tot_err_rate}", file=f)
     return float(tot_err_rate)