de-coder/stt_sw_ctc_model

1---2language:3- sw4license: cc-by-4.05library_name: nemo6datasets:7- mozilla-foundation/common_voice_11_08thumbnail: null9tags:10- automatic-speech-recognition11- speech12- Swahili13- audio14- CTC15- Conformer16- Transformer17- NeMo18- pytorch19model-index:20- name: stt_sw_conformer_ctc_large21  results: []22 23---24 25 26## Model Overview27 28<DESCRIBE IN ONE LINE THE MODEL AND ITS USE>29 30## NVIDIA NeMo: Training31 32To train, fine-tune or play with the model you will need to install [NVIDIA NeMo](https://github.com/NVIDIA/NeMo). We recommend you install it after you've installed latest Pytorch version.33```34pip install nemo_toolkit['all']35``` 36 37## How to Use this Model38 39The model is available for use in the NeMo toolkit [3], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.40 41### Automatically instantiate the model42 43```python44import nemo.collections.asr as nemo_asr45asr_model = nemo_asr.models.ASRModel.from_pretrained("decoder/STT_SW_Model")46```47 48### Transcribing using Python49First, let's get a samples  of  three  different files50```51!wget https://rapidshare.io/2oUF/common_voice_sw_27777611.wav52!wget https://rapidshare.io/2oUG/common_voice_sw_27777627.wav53!wget https://rapidshare.io/2oUH/common_voice_sw_27791961.wav54 55```56Then simply do: if in colab57```58transcriptions = asr_model.transcribe(["/content/common_voice_sw_27777611.wav"])59```60 61### Transcribing many audio files62 63```shell64python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py  pretrained_name="de-coder/stt_rw_conformer_ctc_large"  audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"65```66 67### Input68 69This model accepts 16000 KHz Mono-channel Audio (wav files) as input.70 71### Output72 73This model provides transcribed speech as a string for a given audio sample.74 75## Model Architecture76 77<ADD SOME INFORMATION ABOUT THE ARCHITECTURE>78 79## Training80 81<ADD INFORMATION ABOUT HOW THE MODEL WAS TRAINED - HOW MANY EPOCHS, AMOUNT OF COMPUTE ETC>82 83### Datasets84 85<LIST THE NAME AND SPLITS OF DATASETS USED TO TRAIN THIS MODEL (ALONG WITH LANGUAGE AND ANY ADDITIONAL INFORMATION)>86 87## Performance88 89<LIST THE SCORES OF THE MODEL - 90      OR91USE THE Hugging Face Evaluate LiBRARY TO UPLOAD METRICS>92 93## Limitations94 95<DECLARE ANY POTENTIAL LIMITATIONS OF THE MODEL>96 97Eg: 98Since this model was trained on publically available speech datasets, the performance of this model might degrade for speech which includes technical terms, or vernacular that the model has not been trained on. The model might also perform worse for accented speech.99 100 101## References102 103<ADD ANY REFERENCES HERE AS NEEDED>104 105[1] [NVIDIA NeMo Toolkit](https://github.com/NVIDIA/NeMo)106 107