EncoderDecoderASR model - And how can i cite speechbrain project

Hello
First of all, thank you very much for your amazing job!
I need to use an ASR model for my diploma thesis and I try your pretrained EncoderDecoderASR model from speechbrain.pretrained which works amazingly by the way. I want to know what kind of model it is and on what bibliography it is based. Probably it is an HMM-NN model but can I have more information;
In addition, how can I cite speechbrain project in my thesis; I can’t find out any bibtex .

Hi, glad to hear it’s been helpful!

There are a few models that use that interface; they are all attention-based autoregressive RNN sequence-to-sequence models (not HMM). I don’t know what the best reference for that is; maybe the original attention model paper (“Neural machine translation by jointly learning to align and translate” by D Bahdanau, K Cho, Y Bengio at ICLR 2015).

To cite SpeechBrain, you can use this bibtex entry:

@misc{SB2021,
author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua },
title = {{SpeechBrain}},
year = {2021},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{GitHub - speechbrain/speechbrain: A PyTorch-based Speech Toolkit}},
}

(we’ll be writing a paper about the toolkit soon that you could also cite)

Hello, thank you for your fast reply.
Is there any WER score or any other kind of result for this model?
Is there any recipe to evaluate it myself (I have already downloaded LibriSpeech dataset)?

Hi Alex, you should have a look at our ASR from scratch tutorial :wink: here