You can deploy this model and start earning money today!

xekri/
wav2vec2-common_voice_13_0-eo-10_1

Name: wav2vec2-common_voice_13_0-eo-10_1
Availability: InStock

transformers

pytorch

wav2vec2

automatic-speech-recognition

mozilla-foundation/common_voice_13_0

generated_from_trainer

Views

No views yet

wav2vec2-common_voice_13_0-eo-10_1, an Esperanto speech recognizer

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the mozilla-foundation/common_voice_13_0 Esperanto dataset. It achieves the following results on the evaluation set:

Loss: 0.0391
Cer: 0.0098
Wer: 0.0534

The first 10 examples in the evaluation set:

Actual Predicted	CER
`la orienta parto apud benino kaj niĝerio estis nomita sklavmarbordo` `la orienta parto apud benino kaj niĝerio estis nomita sklafmarbordo`	0.014925373134328358
`en la sekva jaro li ricevis premion` `en la sekva jaro li ricevis premion`	0.0
`ŝi studis historion ĉe la universitato de brita kolumbio` `ŝi studis historion ĉe la universitato de brita kolumbio`	0.0
`larĝaj ŝtupoj kuras al la fasado` `larĝaj ŝtupoj kuras al la fasado`	0.0
`la municipo ĝuas duan epokon de etendo kaj disvolviĝo` `la municipo ĝuas duan epokon de etendo kaj disvolviĝo`	0.0
`li estis ankaŭ katedrestro kaj dekano` `li estis ankaŭ katedresto kaj dekano`	0.02702702702702703
`librovendejo apartenas al la muzeo` `librovendejo apartenas al l muzeo`	0.029411764705882353
`ĝi estas kutime malfacile videbla kaj troviĝas en subkreskaĵaro de arbaroj` `ĝi estas kutime malfacile videbla kaj troviĝas en subkreskaĵo de arbaroj`	0.02702702702702703
`unue ili estas ruĝaj poste brunaj` `unue ili estas ruĝaj poste brunaj`	0.0
`la loĝantaro laboras en la proksima ĉefurbo` `la loĝantaro laboras en la proksima ĉefurbo`	0.0

The differences in results for the above compared to the previous model (xekri/wav2vec2-common_voice_13_0-eo-10) are:

eepokon -> epokon
katedristo -> katedresto
al la muzeo -> al l muzeo

Model description

See facebook/wav2vec2-large-xlsr-53. This model is a version of xekri/wav2vec2-common_voice_13_0-eo-10 trained for 5 more epochs.

Intended uses & limitations

Speech recognition for Esperanto. The base model was pretrained and finetuned on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16KHz.

The output is all lowercase, no punctuation.

Training and evaluation data

The training split was set to train while the eval split was set to validation. Some files were filtered out of the train and validation dataset due to bad data; see xekri/wav2vec2-common_voice_13_0-eo-3 for a detailed discussion. In summary, I used xekri/wav2vec2-common_voice_13_0-eo-3 as a detector to detect bad files, then hardcoded those files into the trainer code to be filtered out.

Training procedure

I used a modified version of run_speech_recognition_ctc.py for training. See run_speech_recognition_ctc.py in this repo.

The parameters to the trainer are in train.json in this repo.

The key changes between this training run and xekri/wav2vec2-common_voice_13_0-eo-3, aside from the filtering and use of the full training and validation sets are:

Layer drop probability is 20%
Train only for 5 epochs

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 3e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
layerdrop: 0.2
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 5
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.1142	0.22	1000	0.0483	0.0126	0.0707
0.1049	0.44	2000	0.0474	0.0123	0.0675
0.0982	0.67	3000	0.0471	0.0120	0.0664
0.092	0.89	4000	0.0459	0.0117	0.0640
0.0847	1.11	5000	0.0459	0.0115	0.0631
0.0837	1.33	6000	0.0453	0.0113	0.0624
0.0803	1.56	7000	0.0443	0.0109	0.0598
0.0826	1.78	8000	0.0441	0.0110	0.0604
0.0809	2.0	9000	0.0437	0.0110	0.0605
0.0728	2.22	10000	0.0451	0.0109	0.0597
0.0707	2.45	11000	0.0444	0.0108	0.0591
0.0698	2.67	12000	0.0442	0.0105	0.0576
0.0981	2.89	13000	0.0411	0.0104	0.0572
0.0928	3.11	14000	0.0413	0.0102	0.0561
0.0927	3.34	15000	0.0410	0.0102	0.0565
0.0886	3.56	16000	0.0402	0.0102	0.0558
0.091	3.78	17000	0.0400	0.0101	0.0553
0.0888	4.0	18000	0.0398	0.0100	0.0546
0.0885	4.23	19000	0.0395	0.0099	0.0542
0.0869	4.45	20000	0.0394	0.0099	0.0540
0.0844	4.67	21000	0.0393	0.0098	0.0539
0.0882	4.89	22000	0.0391	0.0098	0.0537

Framework versions

Transformers 4.29.2
Pytorch 2.0.1+cu117
Datasets 2.12.0
Tokenizers 0.13.3

xekri/wav2vec2-common_voice_13_0-eo-10_1