Vasista22
Rank:Average Model Cost: $0.0000
Number of Runs: 29,881
Models by this creator
whisper-hindi-large-v2
whisper-hindi-large-v2
whisper-hindi-large-v2 is a fine-tuned version of the Whisper ASR model trained on Hindi data from various ASR corpuses. It can be used for speech recognition tasks in Hindi. The model has been trained using the Whisper fine-tuning sprint code and can be evaluated on datasets using the provided evaluation codes. The training data includes the GramVaani ASR Corpus, ULCA ASR Corpus, Shrutilipi ASR Corpus, and Google/Fleurs Train+Dev set, while the evaluation data includes the GramVaani ASR Corpus Test Set and Google/Fleurs Test Set. The model was trained using specific hyperparameters, including a learning rate of 0.75e-05, a training batch size of 8, and a training step of 57000. This work was conducted at the Speech Lab, IIT Madras, with funding from the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
10.3K
Huggingface
whisper-tamil-large-v2
whisper-tamil-large-v2
whisper-tamil-large-v2 is a fine-tuned automatic speech recognition (ASR) model for the Tamil language. It is based on the Whisper ASR model and has been trained on multiple publicly available ASR corpora. The model has been fine-tuned as part of the Whisper fine-tuning sprint. It provides improved speech recognition capabilities for the Tamil language. The model's training data includes various ASR corpora such as IISc-MILE Tamil ASR Corpus, ULCA ASR Corpus, Shrutilipi ASR Corpus, Microsoft Speech Corpus (Indian Languages), Babel ASR Corpus, and Google/Fleurs Train+Dev set. The model's evaluation data includes Microsoft Speech Corpus (Indian Languages) Test Set, Google/Fleurs Test Set, IISc-MILE Test Set, and Babel Test Set. The training hyperparameters include learning rate, batch size, optimizer, seed, learning rate scheduler type, warmup steps, training steps, and mixed precision training. The model's training was done at Speech Lab, IIT Madras, and the compute resources were funded by the "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
8.8K
Huggingface
whisper-kannada-medium
whisper-kannada-medium
The Whisper Kannada Medium model is a fine-tuned version of the openai/whisper-medium model specifically trained for Kannada automatic speech recognition (ASR). It has been trained on multiple publicly available ASR corpora, including the IISc-MILE Kannada ASR Corpus, ULCA ASR Corpus, Shrutilipi ASR Corpus, and Google/Fleurs Train+Dev set. The model's training hyperparameters include a learning rate of 1e-05, a training batch size of 24, an evaluation batch size of 48, and the AdamW optimizer with 8-bit gradients. The training was terminated upon convergence after 13,752 steps. The Whisper Kannada Medium model can be used for ASR tasks in Kannada language.
$-/run
8.5K
Huggingface
whisper-hindi-small
$-/run
1.5K
Huggingface
whisper-telugu-medium
whisper-telugu-medium
Whisper Telugu Medium This model is a fine-tuned version of openai/whisper-medium on the Telugu data available from multiple publicly available ASR corpuses. It has been fine-tuned as a part of the Whisper fine-tuning sprint. NOTE: The code used to train this model is available for re-use in the whisper-finetune repository. Usage In order to evaluate this model on an entire dataset, the evaluation codes available in the whisper-finetune repository can be used. The same repository also provides the scripts for faster inference using whisper-jax. In order to infer a single audio file using this model, the following code snippet can be used: For faster inference of whisper models, the whisper-jax library can be used. Please follow the necessary installation steps as mentioned here, before using the following code snippet: Training and evaluation data Training Data: CSTD IIIT-H ASR Corpus ULCA ASR Corpus Shrutilipi ASR Corpus Microsoft Speech Corpus (Indian Languages) Google/Fleurs Train+Dev set Babel ASR Corpus Evaluation Data: Microsoft Speech Corpus (Indian Languages) Test Set Google/Fleurs Test Set OpenSLR Babel Test Set Training hyperparameters The following hyperparameters were used during training: learning_rate: 1e-05 train_batch_size: 24 eval_batch_size: 48 seed: 22 optimizer: adamw_bnb_8bit lr_scheduler_type: linear lr_scheduler_warmup_steps: 15000 training_steps: 35808 (terminated upon convergence. Initially set to 89520 steps) mixed_precision_training: True Acknowledgement This work was done at Speech Lab, IIT Madras. The compute resources for this work were funded by "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
170
Huggingface
whisper-hindi-medium
$-/run
164
Huggingface
whisper-telugu-base
whisper-telugu-base
Whisper Telugu Base This model is a fine-tuned version of openai/whisper-base on the Telugu data available from multiple publicly available ASR corpuses. It has been fine-tuned as a part of the Whisper fine-tuning sprint. NOTE: The code used to train this model is available for re-use in the whisper-finetune repository. Usage In order to evaluate this model on an entire dataset, the evaluation codes available in the whisper-finetune repository can be used. The same repository also provides the scripts for faster inference using whisper-jax. In order to infer a single audio file using this model, the following code snippet can be used: For faster inference of whisper models, the whisper-jax library can be used. Please follow the necessary installation steps as mentioned here, before using the following code snippet: Training and evaluation data Training Data: CSTD IIIT-H ASR Corpus ULCA ASR Corpus Shrutilipi ASR Corpus Microsoft Speech Corpus (Indian Languages) Google/Fleurs Train+Dev set Babel ASR Corpus Evaluation Data: Microsoft Speech Corpus (Indian Languages) Test Set Google/Fleurs Test Set OpenSLR Babel Test Set Training hyperparameters The following hyperparameters were used during training: learning_rate: 3.3e-05 train_batch_size: 80 eval_batch_size: 88 seed: 22 optimizer: adamw_bnb_8bit lr_scheduler_type: linear lr_scheduler_warmup_steps: 15000 training_steps: 24174 (terminated upon convergence. Initially set to 85952 steps) mixed_precision_training: True Acknowledgement This work was done at Speech Lab, IIT Madras. The compute resources for this work were funded by "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
162
Huggingface
whisper-telugu-large-v2
whisper-telugu-large-v2
Whisper Telugu Large-v2 This model is a fine-tuned version of openai/whisper-large-v2 on the Telugu data available from multiple publicly available ASR corpuses. It has been fine-tuned as a part of the Whisper fine-tuning sprint. NOTE: The code used to train this model is available for re-use in the whisper-finetune repository. Usage In order to evaluate this model on an entire dataset, the evaluation codes available in the whisper-finetune repository can be used. The same repository also provides the scripts for faster inference using whisper-jax. In order to infer a single audio file using this model, the following code snippet can be used: For faster inference of whisper models, the whisper-jax library can be used. Please follow the necessary installation steps as mentioned here, before using the following code snippet: Training and evaluation data Training Data: CSTD IIIT-H ASR Corpus ULCA ASR Corpus Shrutilipi ASR Corpus Microsoft Speech Corpus (Indian Languages) Google/Fleurs Train+Dev set Babel ASR Corpus Evaluation Data: Microsoft Speech Corpus (Indian Languages) Test Set Google/Fleurs Test Set OpenSLR Babel Test Set Training hyperparameters The following hyperparameters were used during training: learning_rate: 0.75e-05 train_batch_size: 8 eval_batch_size: 32 seed: 22 optimizer: adamw_bnb_8bit lr_scheduler_type: linear lr_scheduler_warmup_steps: 22000 training_steps: 75000 mixed_precision_training: True Acknowledgement This work was done at Speech Lab, IIT Madras. The compute resources for this work were funded by "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
125
Huggingface
whisper-tamil-small
whisper-tamil-small
Whisper Tamil Small This model is a fine-tuned version of openai/whisper-small on the Tamil data available from multiple publicly available ASR corpuses. It has been fine-tuned as a part of the Whisper fine-tuning sprint. NOTE: The code used to train this model is available for re-use in the whisper-finetune repository. Usage In order to evaluate this model on an entire dataset, the evaluation codes available in the whisper-finetune repository can be used. The same repository also provides the scripts for faster inference using whisper-jax. In order to infer a single audio file using this model, the following code snippet can be used: For faster inference of whisper models, the whisper-jax library can be used. Please follow the necessary installation steps as mentioned here, before using the following code snippet: Training and evaluation data Training Data: IISc-MILE Tamil ASR Corpus ULCA ASR Corpus Shrutilipi ASR Corpus Microsoft Speech Corpus (Indian Languages) Google/Fleurs Train+Dev set Babel ASR Corpus Evaluation Data: Microsoft Speech Corpus (Indian Languages) Test Set Google/Fleurs Test Set IISc-MILE Test Set Babel Test Set Training hyperparameters The following hyperparameters were used during training: learning_rate: 1.7e-05 train_batch_size: 48 eval_batch_size: 32 seed: 22 optimizer: adamw_bnb_8bit lr_scheduler_type: linear lr_scheduler_warmup_steps: 17500 training_steps: 29659 (Initially set to 84740 steps) mixed_precision_training: True Acknowledgement This work was done at Speech Lab, IIT Madras. The compute resources for this work were funded by "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
123
Huggingface
whisper-telugu-small
whisper-telugu-small
Whisper Telugu Small This model is a fine-tuned version of openai/whisper-small on the Telugu data available from multiple publicly available ASR corpuses. It has been fine-tuned as a part of the Whisper fine-tuning sprint. NOTE: The code used to train this model is available for re-use in the whisper-finetune repository. Usage In order to evaluate this model on an entire dataset, the evaluation codes available in the whisper-finetune repository can be used. The same repository also provides the scripts for faster inference using whisper-jax. In order to infer a single audio file using this model, the following code snippet can be used: For faster inference of whisper models, the whisper-jax library can be used. Please follow the necessary installation steps as mentioned here, before using the following code snippet: Training and evaluation data Training Data: CSTD IIIT-H ASR Corpus ULCA ASR Corpus Shrutilipi ASR Corpus Microsoft Speech Corpus (Indian Languages) Google/Fleurs Train+Dev set Babel ASR Corpus Evaluation Data: Microsoft Speech Corpus (Indian Languages) Test Set Google/Fleurs Test Set OpenSLR Babel Test Set Training hyperparameters The following hyperparameters were used during training: learning_rate: 1.7e-05 train_batch_size: 48 eval_batch_size: 32 seed: 22 optimizer: adamw_bnb_8bit lr_scheduler_type: linear lr_scheduler_warmup_steps: 15000 training_steps: 26856 (terminated upon convergence. Initially set to 89520 steps) mixed_precision_training: True Acknowledgement This work was done at Speech Lab, IIT Madras. The compute resources for this work were funded by "Bhashini: National Language translation Mission" project of the Ministry of Electronics and Information Technology (MeitY), Government of India.
$-/run
89
Huggingface