Releases: TensorSpeech/TensorFlowASR
Releases · TensorSpeech/TensorFlowASR
TensorFlowASR v0.3.0
Release Notes
- Fix conformer tflite on tf-nightly
- Supported Gradients Accumulation for Training in Larger Batch Size
TensorFlowASR v0.2.10
Release Notes
- Reduce differences between
librosa.stftandtf.signal.stft
TensorFlowASR v0.2.9
Release Notes
- Update configurations from dictionary to class
- Update Deep Speech 2
- Supported Jasper
- Add optimizer config to Streaming Transducer Example
TensorFlowASR v0.2.8
Release Notes
- Refactor Streaming RNN Transducer
- Add
projection_unitsoption toTransducerPredictionto use projection layer after layer norm - Fix subword scripts in examples
TensorFlowASR v0.2.7
Release Notes
- Supported Streaming RNN Transducer that uses Unidirectional RNNs as Encoder Layers
- Fix typo in Speech Featurizer
TensorFlowASR v0.2.6
Release Notes
- Move
.txtfiles containing characters to.charactersfiles - Replace
subwords_prefixbysubwordsfor better command-line usage - Add conformer gradient accumulation training example scripts
TensorFlowASR v0.2.5
Release Notes
- Supported Subwords Speech Recognition
- Supported Conformer
- Supported TFLite for Transducer Greedy Decoding