Commit
•
b3b8b68
1
Parent(s):
7213dcd
Correct scripts
Browse files- README.md +5 -4
- get_ctc_tokenizer.py +1 -1
- run_chime4.sh +2 -2
README.md
CHANGED
|
@@ -2,18 +2,19 @@
|
|
| 2 |
language:
|
| 3 |
- en
|
| 4 |
tags:
|
| 5 |
-
-
|
| 6 |
datasets:
|
| 7 |
-
-
|
|
|
|
| 8 |
---
|
| 9 |
|
| 10 |
To reproduce this run, first call `get_ctc_tokenizer.py` to train the CTC tokenizer and then execute the following command to train the CTC system:
|
| 11 |
```python
|
| 12 |
#!/usr/bin/env bash
|
| 13 |
python run_flax_speech_recognition_ctc.py \
|
| 14 |
-
--model_name_or_path="
|
| 15 |
--tokenizer_name="wav2vec2-ctc-chime4-tokenizer" \
|
| 16 |
-
--dataset_name="
|
| 17 |
--dataset_config_name="chime4" \
|
| 18 |
--output_dir="./" \
|
| 19 |
--wandb_project="wav2vec2-ctc" \
|
|
|
|
| 2 |
language:
|
| 3 |
- en
|
| 4 |
tags:
|
| 5 |
+
- esb
|
| 6 |
datasets:
|
| 7 |
+
- esb/datasets
|
| 8 |
+
- ldc/chime-4
|
| 9 |
---
|
| 10 |
|
| 11 |
To reproduce this run, first call `get_ctc_tokenizer.py` to train the CTC tokenizer and then execute the following command to train the CTC system:
|
| 12 |
```python
|
| 13 |
#!/usr/bin/env bash
|
| 14 |
python run_flax_speech_recognition_ctc.py \
|
| 15 |
+
--model_name_or_path="esb/wav2vec2-ctc-pretrained" \
|
| 16 |
--tokenizer_name="wav2vec2-ctc-chime4-tokenizer" \
|
| 17 |
+
--dataset_name="esb/datasets" \
|
| 18 |
--dataset_config_name="chime4" \
|
| 19 |
--output_dir="./" \
|
| 20 |
--wandb_project="wav2vec2-ctc" \
|
get_ctc_tokenizer.py
CHANGED
|
@@ -19,7 +19,7 @@ tokenizer_name = f"wav2vec2-ctc-{dataset_name}-tokenizer"
|
|
| 19 |
cutoff_freq = 0.01
|
| 20 |
|
| 21 |
dataset = load_dataset(
|
| 22 |
-
"
|
| 23 |
dataset_name,
|
| 24 |
split=split,
|
| 25 |
use_auth_token=use_auth_token,
|
|
|
|
| 19 |
cutoff_freq = 0.01
|
| 20 |
|
| 21 |
dataset = load_dataset(
|
| 22 |
+
"esb/datasets",
|
| 23 |
dataset_name,
|
| 24 |
split=split,
|
| 25 |
use_auth_token=use_auth_token,
|
run_chime4.sh
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
#!/usr/bin/env bash
|
| 2 |
python run_flax_speech_recognition_ctc.py \
|
| 3 |
-
--model_name_or_path="
|
| 4 |
--tokenizer_name="wav2vec2-ctc-chime4-tokenizer" \
|
| 5 |
-
--dataset_name="
|
| 6 |
--dataset_config_name="chime4" \
|
| 7 |
--output_dir="./" \
|
| 8 |
--wandb_project="wav2vec2-ctc" \
|
|
|
|
| 1 |
#!/usr/bin/env bash
|
| 2 |
python run_flax_speech_recognition_ctc.py \
|
| 3 |
+
--model_name_or_path="esb/wav2vec2-ctc-pretrained" \
|
| 4 |
--tokenizer_name="wav2vec2-ctc-chime4-tokenizer" \
|
| 5 |
+
--dataset_name="esb/datasets" \
|
| 6 |
--dataset_config_name="chime4" \
|
| 7 |
--output_dir="./" \
|
| 8 |
--wandb_project="wav2vec2-ctc" \
|