Skip to content

Commit dd7216a

Browse files
author
sanchit-gandhi
committed
finish speechbrain
1 parent b994241 commit dd7216a

File tree

8 files changed

+117
-56
lines changed

8 files changed

+117
-56
lines changed

‎speechbrain/run_conformer.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-conformer-transformerlm-librispeech"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=4 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=4 \
2525
--max_eval_samples=-1

‎speechbrain/run_conformersmall.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-conformersmall-transformerlm-librispeech"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=4 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=4 \
2525
--max_eval_samples=-1

‎speechbrain/run_crdnn_rnnlm.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-crdnn-rnnlm-librispeech"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=4 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=4 \
2525
--max_eval_samples=-1

‎speechbrain/run_crdnn_transformerlm.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-crdnn-transformerlm-librispeech"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=4 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=4 \
2525
--max_eval_samples=-1

‎speechbrain/run_eval.py‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -148,7 +148,7 @@ def benchmark(batch):
148148
manifest_path = data_utils.write_manifest(
149149
all_results["references"],
150150
all_results["predictions"],
151-
args.model_id,
151+
args.source,
152152
args.dataset_path,
153153
args.dataset,
154154
args.split,
@@ -186,7 +186,7 @@ def benchmark(batch):
186186
parser.add_argument(
187187
"--dataset_path",
188188
type=str,
189-
default="esb/datasets",
189+
default="hf-audio/esb-datasets-test-only-sorted",
190190
help="Dataset path. By default, it is `esb/datasets`",
191191
)
192192
parser.add_argument(

‎speechbrain/run_transformer_transformerlm.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-transformer-transformerlm-librispeech"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=4 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=4 \
2525
--max_eval_samples=-1

‎speechbrain/run_wav2vec2_commonvoice.sh‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,19 @@ SOURCE="speechbrain/asr-wav2vec2-commonvoice-en"
77
python run_eval.py \
88
--source=$SOURCE \
99
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
10+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
11+
--dataset="librispeech" \
12+
--split="test.clean" \
1313
--device=0 \
1414
--batch_size=16 \
1515
--max_eval_samples=-1
1616

1717
python run_eval.py \
1818
--source=$SOURCE \
1919
--speechbrain_pretrained_class_name="EncoderDecoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
20+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
21+
--dataset="librispeech" \
22+
--split="test.other" \
2323
--device=0 \
2424
--batch_size=16 \
2525
--max_eval_samples=-1

‎speechbrain/run_wav2vec2_librispeech.sh‎

Lines changed: 79 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -3,26 +3,87 @@
33
export PYTHONPATH="..":$PYTHONPATH
44

55
SOURCE="speechbrain/asr-wav2vec2-librispeech"
6+
BATCH_SIZE=32
67

78
python run_eval.py \
8-
--source=$SOURCE \
9-
--speechbrain_pretrained_class_name="EncoderASR" \
10-
--dataset_path="librispeech_asr" \
11-
--dataset="clean" \
12-
--split="test" \
13-
--device=0 \
14-
--batch_size=32 \
15-
--max_eval_samples=-1
16-
17-
python run_eval.py \
18-
--source=$SOURCE \
19-
--speechbrain_pretrained_class_name="EncoderASR" \
20-
--dataset_path="librispeech_asr" \
21-
--dataset="other" \
22-
--split="test" \
23-
--device=0 \
24-
--batch_size=32 \
25-
--max_eval_samples=-1
9+
--source=${SOURCE} \
10+
--speechbrain_pretrained_class_name="EncoderASR" \
11+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
12+
--dataset="ami" \
13+
--split="test" \
14+
--device=0 \
15+
--batch_size=${BATCH_SIZE} \
16+
--max_eval_samples=-1
17+
18+
python run_eval.py \
19+
--source=${SOURCE} \
20+
--speechbrain_pretrained_class_name="EncoderASR" \
21+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
22+
--dataset="earnings22" \
23+
--split="test" \
24+
--device=0 \
25+
--batch_size=${BATCH_SIZE} \
26+
--max_eval_samples=-1
27+
28+
python run_eval.py \
29+
--source=${SOURCE} \
30+
--speechbrain_pretrained_class_name="EncoderASR" \
31+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
32+
--dataset="gigaspeech" \
33+
--split="test" \
34+
--device=0 \
35+
--batch_size=${BATCH_SIZE} \
36+
--max_eval_samples=-1
37+
38+
python run_eval.py \
39+
--source=${SOURCE} \
40+
--speechbrain_pretrained_class_name="EncoderASR" \
41+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
42+
--dataset="librispeech" \
43+
--split="test.clean" \
44+
--device=0 \
45+
--batch_size=${BATCH_SIZE} \
46+
--max_eval_samples=-1
47+
48+
python run_eval.py \
49+
--source=${SOURCE} \
50+
--speechbrain_pretrained_class_name="EncoderASR" \
51+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
52+
--dataset="librispeech" \
53+
--split="test.other" \
54+
--device=0 \
55+
--batch_size=${BATCH_SIZE} \
56+
--max_eval_samples=-1
57+
58+
python run_eval.py \
59+
--source=${SOURCE} \
60+
--speechbrain_pretrained_class_name="EncoderASR" \
61+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
62+
--dataset="spgispeech" \
63+
--split="test" \
64+
--device=0 \
65+
--batch_size=${BATCH_SIZE} \
66+
--max_eval_samples=-1
67+
68+
python run_eval.py \
69+
--source=${SOURCE} \
70+
--speechbrain_pretrained_class_name="EncoderASR" \
71+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
72+
--dataset="tedlium" \
73+
--split="test" \
74+
--device=0 \
75+
--batch_size=${BATCH_SIZE} \
76+
--max_eval_samples=-1
77+
78+
python run_eval.py \
79+
--source=${SOURCE} \
80+
--speechbrain_pretrained_class_name="EncoderASR" \
81+
--dataset_path="hf-audio/esb-datasets-test-only-sorted" \
82+
--dataset="voxpopuli" \
83+
--split="test" \
84+
--device=0 \
85+
--batch_size=${BATCH_SIZE} \
86+
--max_eval_samples=-1
2687

2788
# Evaluate results
2889
RUNDIR=`pwd` && \

0 commit comments

Comments
 (0)