Support computing nbest oracle WER.

egs/librispeech/ASR/conformer_ctc/README.md

@@ -10,15 +10,30 @@ You need to prepare 4 files:
     Supported formats are those supported by `torchaudio.load()`,
     e.g., wav and flac.
 
+Also, you need to install `kaldifeat`. Please refer to
+<https://github.com/csukuangfj/kaldifeat> for installation.
 
-Once you have the above files ready, you can run:
+Once you have the above files ready and have `kaldifeat` installed,
+you can run:
 
 ```
 ./conformer_ctc/pretrained.py \
   --checkpoint /path/to/your/checkpoint.pt \
   --words-file /path/to/words.txt \
   --hlg /path/to/HLG.pt \
-  --sound-file /path/to/your/sound.wav
+  /path/to/your/sound.wav
 ```
 
 and you will see the transcribed result.
+
+If you want to transcribe multiple files at the same time, you can use:
+
+```
+./conformer_ctc/pretrained.py \
+  --checkpoint /path/to/your/checkpoint.pt \
+  --words-file /path/to/words.txt \
+  --hlg /path/to/HLG.pt \
+  /path/to/your/sound1.wav \
+  /path/to/your/sound2.wav \
+  /path/to/your/sound3.wav \
+```

egs/librispeech/ASR/conformer_ctc/pretrained.py

@@ -2,11 +2,15 @@
 
 import argparse
 import logging
+import math
+from typing import List
 
 import k2
+import kaldifeat
 import torch
 import torchaudio
 from conformer import Conformer
+from torch.nn.utils.rnn import pad_sequence
 
 from icefall.decode import get_lattice, one_best_decoding
 from icefall.utils import AttributeDict, get_texts
@@ -38,10 +42,10 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--sound-file",
+        "sound_files",
         type=str,
-        required=True,
-        help="The input sound file to transcribe. "
+        nargs="+",
+        help="The input sound file(s) to transcribe. "
         "Supported formats are those that supported by torchaudio.load(). "
         "For example, wav, flac are supported. "
         "The sample rate has to be 16kHz.",
@@ -56,7 +60,7 @@ def get_params() -> AttributeDict:
             "feature_dim": 80,
             "nhead": 8,
             "num_classes": 5000,
-            "sample_freq": 16000,
+            "sample_rate": 16000,
             "attention_dim": 512,
             "subsampling_factor": 4,
             "num_decoder_layers": 6,
@@ -74,6 +78,30 @@ def get_params() -> AttributeDict:
     return params
 
 
+def read_sound_files(
+    filenames: List[str], expected_sample_rate: float
+) -> List[torch.Tensor]:
+    """Read a list of sound files into a list 1-D float32 torch tensors.
+    Args:
+      filenames:
+        A list of sound filenames.
+      expected_sample_rate:
+        The expected sample rate of the sound files.
+    Returns:
+      Return a list of 1-D float32 torch tensors.
+    """
+    ans = []
+    for f in filenames:
+        wave, sample_rate = torchaudio.load(f)
+        assert sample_rate == expected_sample_rate, (
+            f"expected sample rate: {expected_sample_rate}. "
+            f"Given: {sample_rate}"
+        )
+        # We use only the first channel
+        ans.append(wave[0])
+    return ans
+
+
 def main():
     parser = get_parser()
     args = parser.parse_args()
@@ -87,6 +115,7 @@ def main():
 
     logging.info(f"device: {device}")
 
+    logging.info("Create model")
     model = Conformer(
         num_features=params.feature_dim,
         nhead=params.nhead,
@@ -103,28 +132,39 @@ def main():
     checkpoint = torch.load(args.checkpoint, map_location="cpu")
     model.load_state_dict(checkpoint["model"])
     model.to(device)
+    model.eval()
 
     HLG = k2.Fsa.from_dict(torch.load(params.hlg))
     HLG = HLG.to(device)
 
-    model.to(device)
+    opts = kaldifeat.FbankOptions()
+    opts.device = device
+    opts.frame_opts.dither = 0
+    opts.frame_opts.snip_edges = False
+    opts.frame_opts.samp_freq = params.sample_rate
+    opts.mel_opts.num_bins = params.feature_dim
 
-    wave, sample_freq = torchaudio.load(params.sound_file)
-    assert sample_freq == params.sample_freq
-    wave = wave.to(device)
+    fbank = kaldifeat.Fbank(opts)
 
-    features = torchaudio.compliance.kaldi.fbank(
-        waveform=wave,
-        num_mel_bins=params.feature_dim,
-        snip_edges=False,
-        sample_frequency=params.sample_freq,
+    waves = read_sound_files(
+        filenames=params.sound_files, expected_sample_rate=params.sample_rate
     )
+    waves = [w.to(device) for w in waves]
 
-    features = features.unsqueeze(0)
+    logging.info(f"Decoding started")
+    features = fbank(waves)
 
-    nnet_output, _, _ = model(features)
+    features = pad_sequence(
+        features, batch_first=True, padding_value=math.log(1e-10)
+    )
+
+    with torch.no_grad():
+        nnet_output, _, _ = model(features)
+
+    batch_size = nnet_output.shape[0]
     supervision_segments = torch.tensor(
-        [[0, 0, nnet_output.shape[1]]], dtype=torch.int32
+        [[i, 0, nnet_output.shape[1]] for i in range(batch_size)],
+        dtype=torch.int32,
     )
 
     lattice = get_lattice(
@@ -145,7 +185,14 @@ def main():
     hyps = get_texts(best_path)
     word_sym_table = k2.SymbolTable.from_file(params.words_file)
     hyps = [[word_sym_table[i] for i in ids] for ids in hyps]
-    logging.info(hyps)
+
+    s = "\n"
+    for filename, hyp in zip(params.sound_files, hyps):
+        words = " ".join(hyp)
+        s += f"{filename}:\n{words}\n\n"
+    logging.info(s)
+
+    logging.info(f"Decoding Done")
 
 
 if __name__ == "__main__":