iamjanvijay / rnnt_decoder_cuda
An efficient implementation of RNN-T Prefix Beam Search in C++/CUDA.
☆67Updated 3 years ago
Related projects ⓘ
Alternatives and complementary repositories for rnnt_decoder_cuda
- Memory efficient transducer loss computation☆68Updated 2 years ago
- Recurrent Neural Aligner☆49Updated 4 years ago
- The repo contains our code of ``Semantic Mask for Transformer based End-to-End Speech Recognition"☆37Updated 4 years ago
- An implementation of RNN-Transducer loss in TF-2.0.☆45Updated last year
- ☆75Updated 2 years ago
- mWER loss implementation in tensorflow☆31Updated 4 years ago
- Python wrapper for kaldi's arpa2fst☆38Updated last year
- Dynamic Chunk Streaming and Offline Conformer based on athena-team/Athena.☆43Updated 2 years ago
- A torch implementation of a recursion which turns out to be useful for RNN-T.☆139Updated last year
- ☆65Updated 2 years ago
- Pronunciation-assisted Subword Modeling☆29Updated 5 years ago
- streaming attention networks for end-to-end automatic speech recognition☆55Updated 4 years ago
- ☆48Updated 3 years ago
- Data and code related to the ICASSP submission "A comparison of methods for OOV-word recognition"☆17Updated 2 years ago
- ☆16Updated 2 years ago
- Python wrapper for OpenFST and its extensions from Kaldi. Also support reading/writing ark/scp files☆47Updated 4 months ago
- CMU multilingual speech repository☆31Updated 2 years ago
- Segment a given audio into utterances using a trained end-to-end ASR model.☆73Updated 4 years ago
- Custom decoders for Kaldi☆80Updated 5 years ago
- A pytorch wrapper for LF-MMI training and parallel training in Kaldi☆73Updated 2 years ago
- WIP: pytorch FFI wrapper for Kaldi chain loss (a.k.a. Lattice Free MMI)☆20Updated 5 years ago
- This repository describes our reproducible framework for assessing self-supervised representation learning from speech☆51Updated 3 years ago
- Computes the MWER (minimum WER) Loss with CTC beam search. Knowledge distillation for CTC loss.☆57Updated last year
- HMM, CTC, RNN-Transducer, forward-backward algorithm☆19Updated last year
- it's ASR decoder and make graph project☆32Updated 2 years ago
- Example implementation of Monotonic Chunkwise Attention.☆50Updated 6 years ago
- Conversion of recurrent neural network language models to weighted finite state transducers☆55Updated 6 years ago
- use 3 chinese senteces as training corpus to show how to build lm model and HCLG decoding graph☆8Updated 4 years ago
- An example directory for running Multi-Task Learning training on Kaldi neural networks. In Kaldi-speak, this is an egs dir for nnet3 trai…☆54Updated 4 years ago