whisper-small-vn

ZHProject23/whisper-small-vn

whisper-small-vn is a automatic speech recognition model from ZHProject23. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

2% of all-time downloads

All-time downloads

278

Public

Parameters

242M

52.2 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors967 MB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
License: apache-2.0
Model type: whisper
Access: Public
Created: Dec 14, 2023
Updated: Dec 14, 2023
SHA: af5c4cfb

Base models

openai/whisper-small

Model card

From the Hugging Face model README

Whisper Small Vietnamese

This model is a fine-tuned version of openai/whisper-small on the Vietnamese ASR Custom Corpus dataset. It achieves the following results on the evaluation set:

Loss: 0.2044
Wer: 11.5097

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 8
training_steps: 148

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.5043	0.01	2	4.4639	33.6957
4.0539	0.03	4	3.7975	35.9783
3.3205	0.04	6	3.0084	37.2101
2.7077	0.05	8	2.5555	37.9710
2.2203	0.07	10	2.2051	63.5870
2.1151	0.08	12	1.9006	132.9227
2.0148	0.09	14	1.6122	134.7222
1.2862	0.11	16	1.3244	144.0821
1.207	0.12	18	1.0984	110.8575
1.146	0.14	20	1.0196	105.2053
1.0035	0.15	22	0.9752	99.8792
0.8611	0.16	24	0.9322	80.3986
0.8995	0.18	26	0.8938	69.0700
0.9165	0.19	28	0.8634	55.9058
0.9399	0.2	30	0.8389	44.5894
0.7586	0.22	32	0.8171	38.1884
0.7416	0.23	34	0.7963	34.4444
0.8141	0.24	36	0.7765	31.9082
0.8389	0.26	38	0.7578	34.1546
0.753	0.27	40	0.7393	42.6449
0.756	0.28	42	0.7228	54.4203
0.7347	0.3	44	0.7077	66.2681
0.6526	0.31	46	0.6939	49.2874
0.5208	0.32	48	0.6803	31.5821
0.7804	0.34	50	0.6657	35.3261
0.6831	0.35	52	0.6524	39.2995
0.5237	0.36	54	0.6409	42.6087
0.6799	0.38	56	0.6305	40.0966
0.681	0.39	58	0.6195	44.2754
0.5804	0.41	60	0.6062	42.2101
0.5869	0.42	62	0.5930	44.9517
0.7131	0.43	64	0.5805	41.4734
0.7012	0.45	66	0.5689	42.1498
0.6513	0.46	68	0.5576	41.9203
0.4762	0.47	70	0.5460	43.0193
0.503	0.49	72	0.5333	46.2198
0.5342	0.5	74	0.5200	48.1522
0.4769	0.51	76	0.5055	45.3382
0.3922	0.53	78	0.4910	46.6667
0.4501	0.54	80	0.4752	49.5169
0.4469	0.55	82	0.4574	50.8092
0.5398	0.57	84	0.4356	50.5797
0.3611	0.58	86	0.4053	48.2246
0.4015	0.59	88	0.3565	44.9275
0.3192	0.61	90	0.2724	46.4493
0.2466	0.62	92	0.2516	47.1256
0.2147	0.64	94	0.2574	35.0483
0.1898	0.65	96	0.2431	23.8527
0.2059	0.66	98	0.2314	17.1981
0.2634	0.68	100	0.2258	15.1208
0.1498	0.69	102	0.2223	14.3599
0.2672	0.7	104	0.2202	13.9130
0.0989	0.72	106	0.2192	13.5024
0.2685	0.73	108	0.2181	13.1280
0.1886	0.74	110	0.2168	12.8744
0.3012	0.76	112	0.2155	12.7174
0.2134	0.77	114	0.2143	12.3792
0.3099	0.78	116	0.2133	12.3188
0.336	0.8	118	0.2124	12.2826
0.1386	0.81	120	0.2112	12.0652
0.1756	0.82	122	0.2100	11.9807
0.2789	0.84	124	0.2092	11.8237
0.1284	0.85	126	0.2085	11.7512
0.2586	0.86	128	0.2078	11.6304
0.31	0.88	130	0.2072	11.5942
0.1971	0.89	132	0.2067	11.5700
0.2664	0.91	134	0.2062	11.5459
0.2684	0.92	136	0.2058	11.5459
0.4082	0.93	138	0.2053	11.4976
0.2593	0.95	140	0.2050	11.4855
0.143	0.96	142	0.2048	11.4976
0.2015	0.97	144	0.2046	11.5097
0.1958	0.99	146	0.2045	11.5217
0.3197	1.0	148	0.2044	11.5097

Framework versions

Transformers 4.37.0.dev0
Pytorch 2.0.0+cu117
Datasets 2.15.0
Tokenizers 0.15.0

Task: Automatic Speech Recognition
Library: transformers
Type: whisper
License: apache-2.0
Languages: vi
Created: Dec 14, 2023
Updated: Dec 14, 2023