Automatic Speech Recognition
NeMo
ONNX
GGUF
parakeet
tdt
sherpa-onnx
multilingual
speech-recognition
gabor
fastconformer
Instructions to use aoiandroid/orukeet with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use aoiandroid/orukeet with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("aoiandroid/orukeet") transcriptions = asr_model.transcribe(["file.wav"]) - Notebooks
- Google Colab
- Kaggle
| { | |
| "status": "pass", | |
| "source_sha256": "031c8ddab4845aeced904a7cde8e8aa57993b2e344716cf83a545b079c473b56", | |
| "original_encoder_sha256": "d10711f1b8f3a516e2d7a93adb219caf8aba2b55295305db92f3e80e78c1499a", | |
| "optimized_encoder_sha256": "7b55f2a504a20a8e462899f5befd45f4a1784948d76ed0127902d9cf39405487", | |
| "parent_export_receipt_sha256": "9fc487cdfa076b4d210a526c68ff24ea9ff7ff9da78e058b21f7d610f95ccc6e", | |
| "validation_receipt_sha256": "85295b7612fa7f3dc785c477d7beaae14a90470c7c74911edd15c4d41fdc0aec", | |
| "optimizer_sha256": "d025489a7886020dbbfc5c0f9250f44ae3f372056290b42fa873bc1660e657a1", | |
| "onnx": "1.21.0", | |
| "numpy": "2.4.4", | |
| "method": "24 centered-integer FP32 depthwise Conv substitutions with INT32 outputs", | |
| "replaced_nodes": [ | |
| "/layers.0/conv/depthwise_conv/Conv_quant", | |
| "/layers.1/conv/depthwise_conv/Conv_quant", | |
| "/layers.2/conv/depthwise_conv/Conv_quant", | |
| "/layers.3/conv/depthwise_conv/Conv_quant", | |
| "/layers.4/conv/depthwise_conv/Conv_quant", | |
| "/layers.5/conv/depthwise_conv/Conv_quant", | |
| "/layers.6/conv/depthwise_conv/Conv_quant", | |
| "/layers.7/conv/depthwise_conv/Conv_quant", | |
| "/layers.8/conv/depthwise_conv/Conv_quant", | |
| "/layers.9/conv/depthwise_conv/Conv_quant", | |
| "/layers.10/conv/depthwise_conv/Conv_quant", | |
| "/layers.11/conv/depthwise_conv/Conv_quant", | |
| "/layers.12/conv/depthwise_conv/Conv_quant", | |
| "/layers.13/conv/depthwise_conv/Conv_quant", | |
| "/layers.14/conv/depthwise_conv/Conv_quant", | |
| "/layers.15/conv/depthwise_conv/Conv_quant", | |
| "/layers.16/conv/depthwise_conv/Conv_quant", | |
| "/layers.17/conv/depthwise_conv/Conv_quant", | |
| "/layers.18/conv/depthwise_conv/Conv_quant", | |
| "/layers.19/conv/depthwise_conv/Conv_quant", | |
| "/layers.20/conv/depthwise_conv/Conv_quant", | |
| "/layers.21/conv/depthwise_conv/Conv_quant", | |
| "/layers.22/conv/depthwise_conv/Conv_quant", | |
| "/layers.23/conv/depthwise_conv/Conv_quant" | |
| ], | |
| "taps_per_kernel": 9, | |
| "integer_product_magnitude_bound": 65025, | |
| "integer_partial_sum_magnitude_bound": 585225, | |
| "fp32_exact_integer_limit": 16777216, | |
| "original_quantized_initializers_retained": true, | |
| "dynamic_quantization_and_scales_unchanged": true, | |
| "model_metadata_io_and_opsets_unchanged": true, | |
| "new_runtime_dependencies": [], | |
| "validation": { | |
| "per_operator_cases": 72, | |
| "full_encoder_cases": 4, | |
| "all_bit_exact": true | |
| }, | |
| "optimized_graph": { | |
| "file": "encoder.int8.onnx", | |
| "bytes": 653182378, | |
| "sha256": "7b55f2a504a20a8e462899f5befd45f4a1784948d76ed0127902d9cf39405487", | |
| "ir_version": 8, | |
| "opset_imports": { | |
| "ai.onnx": 17 | |
| }, | |
| "inputs": [ | |
| { | |
| "name": "audio_signal", | |
| "type": "FLOAT, audio_signal_dynamic_axes_1x128xaudio_signal_dynamic_axes_2" | |
| }, | |
| { | |
| "name": "length", | |
| "type": "INT64, length_dynamic_axes_1" | |
| } | |
| ], | |
| "outputs": [ | |
| { | |
| "name": "outputs", | |
| "type": "FLOAT, Transposeoutputs_dim_0x1024xTransposeoutputs_dim_2" | |
| }, | |
| { | |
| "name": "encoded_lengths", | |
| "type": "INT64, length_dynamic_axes_1" | |
| } | |
| ], | |
| "metadata": { | |
| "vocab_size": "8192", | |
| "normalize_type": "per_feature", | |
| "pred_rnn_layers": "2", | |
| "pred_hidden": "640", | |
| "subsampling_factor": "8", | |
| "model_type": "EncDecRNNTBPEModel", | |
| "version": "2", | |
| "model_author": "NeMo", | |
| "url": "https://huggingface.co/oruk/orukeet#parakeet-tdt-v3", | |
| "comment": "Orukeet r3; standard Parakeet TDT v3 transducer graph", | |
| "feat_dim": "128", | |
| "orukeet_source_sha256": "031c8ddab4845aeced904a7cde8e8aa57993b2e344716cf83a545b079c473b56" | |
| }, | |
| "operators": [ | |
| "ai.onnx::Add", | |
| "ai.onnx::And", | |
| "ai.onnx::Cast", | |
| "ai.onnx::Concat", | |
| "ai.onnx::Constant", | |
| "ai.onnx::ConstantOfShape", | |
| "ai.onnx::Conv", | |
| "ai.onnx::ConvInteger", | |
| "ai.onnx::Div", | |
| "ai.onnx::DynamicQuantizeLinear", | |
| "ai.onnx::Equal", | |
| "ai.onnx::Expand", | |
| "ai.onnx::Gather", | |
| "ai.onnx::LayerNormalization", | |
| "ai.onnx::Less", | |
| "ai.onnx::MatMul", | |
| "ai.onnx::MatMulInteger", | |
| "ai.onnx::Mul", | |
| "ai.onnx::Not", | |
| "ai.onnx::Pad", | |
| "ai.onnx::Range", | |
| "ai.onnx::Relu", | |
| "ai.onnx::Reshape", | |
| "ai.onnx::Shape", | |
| "ai.onnx::Sigmoid", | |
| "ai.onnx::Slice", | |
| "ai.onnx::Softmax", | |
| "ai.onnx::Split", | |
| "ai.onnx::Squeeze", | |
| "ai.onnx::Sub", | |
| "ai.onnx::Tile", | |
| "ai.onnx::Transpose", | |
| "ai.onnx::Unsqueeze", | |
| "ai.onnx::Where" | |
| ], | |
| "external_files": [] | |
| } | |
| } | |