Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition

Franck Dernoncourt; Petr Motlicek; Subhadeep Dey; Trung Bui

REVIEW 4 major objections 5 minor 29 references

Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read By leaving dropout on during decoding, a semi-supervised speech recogniser generates multiple candidate transcripts of each unlabelled utterance, filters the confident ones, and retrains to cut word error rate by 2% absolute.

desk verdict A plausible semi-supervised ASR trick with one headline number that is currently confounded by adding both extra hypotheses and a different decoder. read the letter →

arxiv 1908.05227 v1 pith:LRU45WTJ submitted 2019-08-08 eess.AS cs.LGcs.SDstat.ML

Subhadeep Dey , Petr Motlicek , Trung Bui , Franck Dernoncourt This is my paper · ORCID

classification eess.AScs.LGcs.SDstat.ML

keywords semi-supervisedlearningend-to-endspeechrecognitiondropoutdataselectionN-besthypothesesworderrorrateself-trainingconfidenceestimation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

End-to-end speech recognisers need large labelled corpora, but labelled audio is expensive and unlabelled audio is cheap. This paper tests a way to exploit unlabelled audio: keep dropout, the training-time trick of randomly switching off network units, active during decoding so each utterance yields several candidate transcripts, then keep only the high-confidence transcripts and retrain on them. On a public lecture-transcription corpus this loop lowered word error rate from 29.2% to 27.2%, a 2% absolute improvement over supervised adaptation alone. The paper also shows the recipe is fragile: on a proprietary command dataset, when the confidence threshold was borrowed from another domain, the dropout-augmented transcripts made errors worse, 34.3% versus 32.2%.

What carries the argument

The central mechanism is the dropout-augmented N-best hypothesis set. The seed model is an encoder-decoder network with attention whose training objective also includes connectionist temporal classification (CTC) loss, and it is trained with dropout, the standard regulariser that randomly turns off units. At inference the dropout is deliberately left on, so decoding the same utterance several times produces several plausible transcripts instead of one; these transcripts are added to the ordinary N-best list. A confidence filter, either the beam-search decoding score or the per-utterance entropy of character posteriors, then admits selected hypotheses into the retraining data, and the selected transcripts retrain the seed model. The dropout pass is the workhorse: it converts a single point estimate into a small stochastic ensemble that reveals where the model is uncertain.

What would settle it

On a new domain with a same-domain development set for threshold calibration, compare retraining with dropout-augmented transcripts against retraining with ordinary 10-best transcripts on the same selected utterances; if the dropout-augmented model does not achieve a lower word error rate, the proposed mechanism fails.

Watch

Extended reading notes

Core claim

The discovery is that a single confidence-filtered hypothesis underuses an end-to-end ASR model in semi-supervised training; sampling several hypotheses by keeping dropout active at inference exposes where the model is uncertain and gives the data-selection stage better material to filter. A seed model trained on source speech and adapted to 15 hours of labelled target speech is used to decode 50 hours of unlabelled target audio. Applying dropout during these decoding passes to build an augmented N-best list, then selecting utterances whose decoding score passed a threshold of -0.5, produced a retrained model with 27.2% word error rate on the target test set, against 29.2% for the supervised-only model and 28.9% when the same selection was applied to ordinary 10-best lists. The paper reports that entropy-based selection gave smaller or no gains, and that dropout augmentation can reverse its benefit when the confidence threshold does not transfer.

Load-bearing premise

The confidence scores computed from the seed model reliably identify which automatically generated transcripts are accurate enough to serve as training labels.

Editorial extensions

If this is right

Retraining on dropout-augmented, confidence-filtered hypotheses is a plug-in addition: it needs no new labelled data, no architectural change, and no external text, only several decoding passes per utterance.
The 2% gain over supervised adaptation on the lecture corpus indicates that the value comes from hypothesis diversity, not from the selection step alone, because the same selection with ordinary 10-best transcripts gave only a 0.3% gain.
Entropy-based utterance confidence is a weaker selector than the decoding score in this setup, so practitioners should prefer score thresholds for data selection when they are available.
The cross-domain result is a calibration warning: dropout augmentation should be paired with confidence thresholds tuned on the target domain, or it can degrade performance.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural extension is to average the output distributions over the dropout passes rather than picking discrete transcripts, which would preserve more of the uncertainty information for the selection stage.
The same loop could transfer to other sequence-generation tasks that already use dropout, such as handwriting recognition or speech translation.
Because the proprietary-dataset failure is attributed to a borrowed threshold, a testable fix is to reserve a small labelled development set in the target domain purely for threshold calibration; the paper did this for the lecture corpus but not for the command dataset.
The reported 2% gain rests on a single test set; averaging over random seeds and repeated decoding passes would show whether the gain is stable.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a small, plausible extension of dropout-as-uncertainty to semi-supervised end-to-end ASR, and the one clean-looking result (2% absolute WER gain on TEDLIUM) is not actually clean. I'd send it to review, but the referee should ask for a matched control before believing the dropout story.

The original bit is the combination: use MC-dropout at inference to generate multiple 1-best transcripts, pool them with the seed model's N-best list, filter by confidence, and retrain. That is a legitimate extension of Gal & Ghahramani and of confidence-based self-training, and I don't know of a prior end-to-end ASR paper that does exactly this. The paper also deserves credit for reporting the Adobe internal dataset failure: adding the dropout transcripts made WER worse (34.3 vs 32.2), and the text says this might be a threshold issue without testing it. Disappointing but honest.

The soft spot is the TEDLIUM comparison. E2ETED S+U retrains on the seed model's 10-best; E2ETED S+U + E2E-drop retrains on those same 10-best plus extra transcripts produced by a separately trained non-adapted dropout model. That changes two things at once: the number and the source of the pseudo-labels. There is no condition that adds, say, more 10-best hypotheses from E2ES to match the count, so the 1.7% gain cannot be pinned on dropout's uncertainty localization. It might just be that more hypotheses help, or that a different model's transcripts add diversity. The paper even notes E2ES N-best beyond 10 didn't help, but that is not a matched control. The number of dropout repetitions N is never reported. And the abstract says 'significantly' without any significance test or error bars. The Adobe result actually undermines the dropout-specific claim further, because there the extra dropout hypotheses hurt.

Entropy-based selection also does essentially nothing on TEDLIUM (29.2 baseline vs 28.8 with dropout). So the evidence for the core mechanism is thin, though the overall approach of confidence-filtered self-training with N-best lists is fine and gives a small gain on one dataset.

Who is this for? People tuning semi-supervised end-to-end ASR in low-resource settings may want to try this, but I wouldn't change my system based on the current numbers. It deserves a serious referee: the idea is plausible, the write-up is clear, and the failure case is reported. The referee should first ask for a matched control and the missing N value, then decide if the dropout story is still standing. My own verdict would be conditional, not reject.

Referee Report

4 major / 5 minor

Summary. The paper proposes a semi-supervised training scheme for end-to-end speech recognition. A seed end-to-end model is first trained on LibriSpeech and adapted to a small labeled target-domain set; it is then used to decode unlabeled target-domain audio, and the resulting high-confidence transcripts are selected by either a decoding-score threshold or an utterance-level entropy criterion and used to retrain the model. To capture model uncertainty, the authors additionally apply dropout at inference time to generate multiple one-best transcripts, which are appended to the N-best list used for retraining. Experiments are reported on TEDLIUM and on an internal Adobe dataset. The best reported result is a TEDLIUM WER of 27.2% for the dropout-augmented semi-supervised system versus 29.2% for the supervised-only adapted end-to-end system and 28.9% for the semi-supervised system without dropout. On the Adobe dataset, however, the dropout-augmented variant is worse than the non-dropout semi-supervised system when using decoding-score selection (34.3% vs. 32.2%).

Significance. If the central claim were cleanly supported, the paper would make a modest but useful contribution: it connects Bayesian-dropout-style uncertainty estimation with self-training for end-to-end ASR, and it reports experiments on two domains. The paper also honestly reports a negative transfer result on the Adobe dataset. However, the current experimental design does not isolate the effect of dropout: the decisive TEDLIUM comparison changes the number of pseudo-labels, the generating model, and the inference procedure at the same time. In addition, no uncertainty estimates or significance tests are provided, and the entropy-based selection procedure is not fully specified. Thus the paper's significance is contingent on additional controlled experiments, which are within the scope of a revision.

major comments (4)

[Section 5.2.1, Table 4] The decisive TEDLIUM comparison is confounded. E2ETED S+U is retrained on the 10-best hypotheses produced by E2ES, while E2ETED S+U + E2E-drop is retrained on those same hypotheses plus an unspecified number of 1-best transcripts obtained by applying dropout to E2E-drop, a separately trained source-domain model that is not adapted to TEDLIUM and has worse source-domain performance than E2E (39.1% vs. 38.5% in Table 2). The treatment therefore changes three variables at once: the number of pseudo-labels per utterance, the model that generated them, and whether dropout is applied at inference. There is no control condition that adds an equal number of additional hypotheses from E2ES without dropout. The statement in Section 3 that dropout diversity 'will implicitly increase reliability' is a hypothesis, not evidence. The reported gain from 28.9% to 27.2% cannot be attributed to dropout unless such a control is provided, and the number of dropout repetitions N in Section 3 is never reported, so the treatment magnitude is also unspecified.
[Section 5.3, Table 4] The Adobe experiment contradicts the general claim that the proposed approach reduces ASR errors. With decoding-score selection, E2EAb U + E2E-drop yields 34.3% WER, which is 2.1% absolute worse than E2EAb U at 32.2%; with entropy selection the difference is 37.7% vs. 38.1%, far smaller than the TEDLIUM gain. The paper attributes the failure to a non-optimal threshold, but no experiment varies the threshold on Adobe data, and no Adobe development set is available to test this explanation. The abstract's statement that experiments on Adobe data show a reduction in ASR errors is therefore not supported by the table. The authors should either provide a matched threshold analysis, or restrict the claim to TEDLIUM and state the transfer limitation explicitly.
[Section 5, Tables 3 and 4] No uncertainty quantification is provided for any of the reported WER differences. The TEDLIUM and Adobe test sets are only 2.5 hours each, and the claimed absolute gains are 0.3%, 0.4%, 1.7%, and 2.0%. Without error bars, confidence intervals, or significance tests, the abstract's phrase 'significantly reduces ASR errors' cannot be verified. Please report, for example, bootstrap confidence intervals over utterances or results across multiple test-set partitions.
[Section 3, Eq. (2), Section 5.2.2] The entropy-based data-selection procedure is underspecified. Step 2 of the proposed algorithm says to select an utterance if 'the entropy or the decoding-score is above a threshold,' but for entropy a low value is the usual indicator of confidence, so the selection direction should be clarified. The threshold value used for the entropy criterion in Table 4 is not reported, and Eq. (2) is ambiguous: it is not clear whether c indexes the character vocabulary and T is the number of time frames, or whether p(c|X) is a per-frame posterior. Without this information, the entropy results cannot be reproduced.

minor comments (5)

[Abstract] The phrase 'an speech recording' should be corrected to 'a speech recording'.
[Section 4.2] The sentence 'we use 3 hours and 50 hours of data as the cross validation and unsupervised set (dev3 from Table 1) respectively' is ambiguous because Table 1 lists dev3 as 3 hours and dev2 as 50 hours; the intended mapping between dev2/dev3 and the two roles should be stated explicitly.
[Section 2] The term 'connectionalist temporal classification' should be 'connectionist temporal classification'.
[Section 5.2.1] The claim that thresholds less than -0.3 lead to the selection of shorter-duration utterances is not accompanied by any supporting figure or table; please add evidence or soften the statement.
[Figure 2] The histogram would benefit from axis labels and a caption explaining how the false-alarm and miss-detection trade-off was read from the plot.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the claimed gain is a measured test-set WER, and no step reduces to its inputs by construction.

full rationale

The paper's central claim is an empirical result: the proposed E2ETED S+U + E2E-drop system achieves 27.2% WER on the TEDLIUM test set versus 29.2% for the supervised E2ES baseline. This outcome is evaluated on a held-out test set and is not derived from the confidence scores or the dropout mechanism by construction. The decoding-score threshold is chosen on the TEDLIUM dev3 cross-validation data and then applied to the dev2 unsupervised set, which is a standard experimental procedure rather than a fitted parameter presented as a prediction. The dropout-augmented transcripts are generated by a separately trained E2E-drop model, and whether adding them helps retraining is an empirical question answered by the test-set result. The paper explicitly labels the reliability benefit as a hypothesis ('We assume that the diversity of automatically generated transcripts for an utterance will implicitly increase the reliability of the model'), not as a theorem or a derivation. No load-bearing self-citation, imported uniqueness theorem, or ansatz smuggled in via citation is used to force the conclusion. The Adobe negative result (34.3% with dropout versus 32.2% without) is a robustness failure and a possible experimental confound, but it is not circular reasoning. Therefore, no circular step can be exhibited, and the appropriate score is 0.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical or model entities. The load-bearing ingredients are the confidence thresholds, the N-best size, the dropout sampling repetition count, and the domain assumptions about dropout uncertainty and confidence-reliability correlation. The threshold and N-best size are fitted to development data; the dropout repetition count is not even reported.

free parameters (5)

decoding-score threshold = -0.5
Tuned on TEDLIUM dev3 by inspecting histograms of decoding scores for utterances with WER <=10% vs >10% (Section 5.2.1, Figure 2), then reused on Adobe without re-tuning.
N-best list size = 10
Authors state 10-best helps more than 1-best and that applying more than 10-best gives no further improvement (Section 5.2.1); this is an empirically chosen hyperparameter.
number of dropout decoding repetitions N = not specified
The algorithm repeats dropout decoding N times (Section 3, step 3) but the paper never reports the value of N used in the experiments, which directly controls how many augmented transcripts are generated.
dropout probability = 0.2
The dropout-augmented model E2E-drop applies dropout 0.2 to all encoder and decoder layers (Section 5); this value is chosen by hand and affects diversity of sampled transcripts.
entropy selection threshold = not stated
The entropy-based data selection (Section 5.2.2) is applied without reporting the threshold value or how it was chosen, so the comparison is underspecified.

assumptions (3)

domain assumption Dropout applied at inference time approximates sampling from the model's posterior weight distribution and thus captures model uncertainty.
The method is motivated by Gal and Ghahramani [15] and Vyas et al. [14]; the paper assumes diversity of dropout-generated transcripts localizes uncertainty (Sections 1 and 3).
domain assumption The decoding-score and entropy confidence measures correlate with transcript correctness on unlabeled data.
Figure 2 shows a heuristic separation on TEDLIUM dev3, but the paper does not establish a calibrated relationship, especially for Adobe where the threshold is transferred (Section 5.3).
domain assumption The end-to-end ASR trained on 100h LibriSpeech can serve as a useful seed after adaptation to 15h target labeled data.
Section 4.1 and 4.2; the seed model's quality limits the quality of generated transcripts used in self-training.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition." pith.science (2026). https://pith.science/paper/LRU45WTJ

@misc{pith2026190805227,
  author       = {Pith},
  title        = {Pith review of: Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LRU45WTJ}},
  note         = {Machine review of arXiv:1908.05227}
}

read the original abstract

In this paper, we explore various approaches for semi supervised learning in an end to end automatic speech recognition (ASR) framework. The first step in our approach involves training a seed model on the limited amount of labelled data. Additional unlabelled speech data is employed through a data selection mechanism to obtain the best hypothesized output, further used to retrain the seed model. However, uncertainties of the model may not be well captured with a single hypothesis. As opposed to this technique, we apply a dropout mechanism to capture the uncertainty by obtaining multiple hypothesized text transcripts of an speech recording. We assume that the diversity of automatically generated transcripts for an utterance will implicitly increase the reliability of the model. Finally, the data selection process is also applied on these hypothesized transcripts to reduce the uncertainty. Experiments on freely available TEDLIUM corpus and proprietary Adobe's internal dataset show that the proposed approach significantly reduces ASR errors, compared to the baseline model.

Figures

Figures reproduced from arXiv: 1908.05227 by the authors.

**Figure 1.** Architecture for encoder decoder network for end-toend ASR. 2. End-to-end ASR The state-of-the-art based ASR (chain model) requires a lexicon and alignments, usually generated with respect to contextdependent tri-phonetic states [3]. An alternative technique, referred to as end-to-end, aims to learn the mapping from acoustic features to text directly without the need of intermediate steps. Recently, various end-… view at source ↗

**Figure 2.** Histogram plot of decoding-scores for set of utterances, with WER ≤ 10% (blue) and decoding-scores for utterances with WER > 10% (orange). 6. Conclusions Techniques for semi-supervised learning for ASR were investigated in this paper. For exploiting unlabelled data, the baseline system employs a single best hypothesized text-transcript. As opposed to this approach, we proposed to capture the uncertainties by ap… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 24 canonical work pages

[1]

Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition

Introduction State-of-the-art approaches in automatic speech recognition (ASR) exploit the powerful discriminative capability of deep neural networks (DNN) for acoustic modelling [1, 2, 3]. The current ASR advancements offer low error-rates, making the systems applicable for commercialization. In the past few years, sequence level optimization algorithms,...

work page Pith review arXiv 1908
[2]

An alternative technique, re- ferred to as end-to-end, aims to learn the mapping from acoustic features to text directly without the need of intermediate steps

End-to-end ASR The state-of-the-art based ASR (chain model) requires a lexi- con and alignments, usually generated with respect to context- dependent tri-phonetic states [3]. An alternative technique, re- ferred to as end-to-end, aims to learn the mapping from acoustic features to text directly without the need of intermediate steps. Recently, various end...

work page
[3]

Semi supervised learning The end-to-end ASR is typically trained with a large amount (at least∼100 hours) of labelled data [16]. However in a semi- supervised setting, it is assumed that only a small amount of supervised data (∼10 to 15 hours) is available for training in addition to a large amount of untranscribed audio for the target domain. Estimating ...

work page
[4]

Dropout: apply dropout during the inference to obtain 1-best hypothesized transcript

work page
[5]

Data-selection: augment the adaptation data with this ut- terance if the entropy or the decoding-score is above a threshold

work page
[6]

Repeat steps 1 and 2 for N times The above steps are applied to all the utterances of the un- supervised dataset

work page
[7]

Experiments are performed on TEDLIUM and Table 1: Training, adaptation and test data for different dataset

Experimental Setup In this section, the experimental setup for the semi-supervised ASR is detailed. Experiments are performed on TEDLIUM and Table 1: Training, adaptation and test data for different dataset. The dev1 represents the supervised data while dev2 comprises the unlabelled-data. Data LibriSpeech TEDLIUM Adobe train 100 hours - - dev1 - 15 hours ...

work page
[8]

The following ASR systems will be analyzed: • LF-MMI: This ASR refers to the traditional chain model using LF-MMI optimization criteria

Results In this section, the results of the end-to-end ASR are presented. The following ASR systems will be analyzed: • LF-MMI: This ASR refers to the traditional chain model using LF-MMI optimization criteria. The system is described in Section 4.4. This system is trained using the standard kaldi’s recipe [19]. • End-to-end: This refers to end-to-end ASR...

work page

Show all 29 references

[9]

For exploiting unlabelled data, the base- line system employs a single best hypothesized text-transcript

Conclusions Techniques for semi-supervised learning for ASR were inves- tigated in this paper. For exploiting unlabelled data, the base- line system employs a single best hypothesized text-transcript. As opposed to this approach, we proposed to capture the un- certainties by a...
[10]

SM2 - Extracting semantic meaning from spoken material

Acknowledgements This work was done under the “SM2 - Extracting semantic meaning from spoken material” project, partially supported by the Swiss Innovation Agency (InnoSuisse) as well as by a re- search grant from Adobe Research, USA
[11]

Deep neural networks for acoustic modeling in speech recognition,

G. Hinton, L. Deng, D. Yu, G. Dahl, A.-r. Mohamed, N. Jaitly, A. Senior, V . Vanhoucke, P. Nguyen, B. Kingsburyet al., “Deep neural networks for acoustic modeling in speech recognition,” IEEE Signal processing magazine, vol. 29, 2012

2012
[12]

New types of deep neural network learning for speech recognition and related applications: An overview,

L. Deng, G. Hinton, and B. Kingsbury, “New types of deep neural network learning for speech recognition and related applications: An overview,” in2013 IEEE International Conference on Acous- tics, Speech and Signal Processing. IEEE, 2013, pp. 8599–8603

2013
[13]

Purely sequence-trained neu- ral networks for asr based on lattice-free mmi

D. Povey, V . Peddinti, D. Galvez, P. Ghahremani, V . Manohar, X. Na, Y . Wang, and S. Khudanpur, “Purely sequence-trained neu- ral networks for asr based on lattice-free mmi.” 2016

2016
[14]

Exploiting foreign resources for dnn-based asr,

P. Motlicek, D. Imseng, B. Potard, P. N. Garner, and I. Himawan, “Exploiting foreign resources for dnn-based asr,”EURASIP Jour- nal on Audio, Speech, and Music Processing , vol. 2015, no. 1, p. 17, 2015

2015
[15]

Joint ctc-attention based end-to-end speech recognition using multi-task learning,

S. Kim, T. Hori, and S. Watanabe, “Joint ctc-attention based end-to-end speech recognition using multi-task learning,” in2017 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2017, pp. 4835–4839

2017
[16]

Espnet: End-to-end speech processing toolkit,

S. Watanabe, T. Hori, S. Karita, T. Hayashi, J. Nishitoba, Y . Unno, N. E. Y . Soplin, J. Heymann, M. Wiesner, N. Chen et al. , “Espnet: End-to-end speech processing toolkit,” arXiv preprint arXiv:1804.00015, 2018

2018 arXiv
[17]

Eesen: End-to-end speech recognition using deep rnn models and wfst-based decoding,

Y . Miao, M. Gowayyed, and F. Metze, “Eesen: End-to-end speech recognition using deep rnn models and wfst-based decoding,” in 2015 IEEE Workshop on Automatic Speech Recognition and Un- derstanding (ASRU). IEEE, 2015, pp. 167–174

2015
[18]

Semi- supervised training of acoustic models using lattice-free mmi,

V . Manohar, H. Hadian, D. Povey, and S. Khudanpur, “Semi- supervised training of acoustic models using lattice-free mmi,” in 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018, pp. 4844–4848

2018
[19]

Semi-supervised end-to-end speech recognition,

S. Karita, S. Watanabe, T. Iwata, A. Ogawa, and M. Delcroix, “Semi-supervised end-to-end speech recognition,” in Proc. Inter- speech, 2018, pp. 2–6

2018
[20]

Pseudo-label: The simple and efﬁcient semi- supervised learning method for deep neural networks,

D.-H. Lee, “Pseudo-label: The simple and efﬁcient semi- supervised learning method for deep neural networks,” in Work- shop on Challenges in Representation Learning, ICML , vol. 3, 2013, p. 2

2013
[21]

Semi-supervised model training for unbounded conversational speech recognition,

S. Walker, M. Pedersen, I. Orife, and J. Flaks, “Semi-supervised model training for unbounded conversational speech recognition,” arXiv preprint arXiv:1705.09724, 2017

2017 arXiv
[22]

Learning with pseudo- ensembles,

P. Bachman, O. Alsharif, and D. Precup, “Learning with pseudo- ensembles,” in Advances in Neural Information Processing Sys- tems, 2014, pp. 3365–3373

2014
[23]

Dropout: A simple way to prevent neural networks from overﬁtting,

N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, “Dropout: A simple way to prevent neural networks from overﬁtting,” Journal of Machine Learning Research, vol. 15, pp. 1929–1958, 2014. [Online]. Available: http://jmlr.org/papers/v15/srivastava14a.html

1929
[24]

Analyzing uncer- tainties in speech recognition using dropout,

A. Vyas, P. Dighe, S. Tong, and H. Bourlard, “Analyzing uncer- tainties in speech recognition using dropout,” in Proceedings of the IEEE International Conference on Acoustics, Speech, and Sig- nal Processing (ICASSP), Jul. 2019

2019
[25]

Dropout as a bayesian approxima- tion: Representing model uncertainty in deep learning,

Y . Gal and Z. Ghahramani, “Dropout as a bayesian approxima- tion: Representing model uncertainty in deep learning,” in inter- national conference on machine learning , 2016, pp. 1050–1059

2016
[26]

End-to-end speech recogni- tion with word-based rnn language models,

T. Hori, J. Cho, and S. Watanabe, “End-to-end speech recogni- tion with word-based rnn language models,” in 2018 IEEE Spo- ken Language Technology Workshop (SLT) . IEEE, 2018, pp. 389–396

2018
[27]

Lib- rispeech: an asr corpus based on public domain audio books,

V . Panayotov, G. Chen, D. Povey, and S. Khudanpur, “Lib- rispeech: an asr corpus based on public domain audio books,” in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2015, pp. 5206–5210

2015
[28]

Ted-lium: an automatic speech recognition dedicated corpus

A. Rousseau, P. Del ´eglise, and Y . Esteve, “Ted-lium: an automatic speech recognition dedicated corpus.” in LREC, 2012, pp. 125– 129

2012
[29]

The kaldi speech recognition toolkit,

D. Povey, A. Ghoshal, G. Boulianne, L. Burget, O. Glembek, N. Goel, M. Hannemann, P. Motlicek, Y . Qian, P. Schwarzet al., “The kaldi speech recognition toolkit,” IEEE Signal Processing Society, Tech. Rep., 2011

2011

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition

Introduction State-of-the-art approaches in automatic speech recognition (ASR) exploit the powerful discriminative capability of deep neural networks (DNN) for acoustic modelling [1, 2, 3]. The current ASR advancements offer low error-rates, making the systems applicable for commercialization. In the past few years, sequence level optimization algorithms,...

work page Pith review arXiv 1908

[2] [2]

An alternative technique, re- ferred to as end-to-end, aims to learn the mapping from acoustic features to text directly without the need of intermediate steps

End-to-end ASR The state-of-the-art based ASR (chain model) requires a lexi- con and alignments, usually generated with respect to context- dependent tri-phonetic states [3]. An alternative technique, re- ferred to as end-to-end, aims to learn the mapping from acoustic features to text directly without the need of intermediate steps. Recently, various end...

work page

[3] [3]

Semi supervised learning The end-to-end ASR is typically trained with a large amount (at least∼100 hours) of labelled data [16]. However in a semi- supervised setting, it is assumed that only a small amount of supervised data (∼10 to 15 hours) is available for training in addition to a large amount of untranscribed audio for the target domain. Estimating ...

work page

[4] [4]

Dropout: apply dropout during the inference to obtain 1-best hypothesized transcript

work page

[5] [5]

Data-selection: augment the adaptation data with this ut- terance if the entropy or the decoding-score is above a threshold

work page

[6] [6]

Repeat steps 1 and 2 for N times The above steps are applied to all the utterances of the un- supervised dataset

work page

[7] [7]

Experiments are performed on TEDLIUM and Table 1: Training, adaptation and test data for different dataset

Experimental Setup In this section, the experimental setup for the semi-supervised ASR is detailed. Experiments are performed on TEDLIUM and Table 1: Training, adaptation and test data for different dataset. The dev1 represents the supervised data while dev2 comprises the unlabelled-data. Data LibriSpeech TEDLIUM Adobe train 100 hours - - dev1 - 15 hours ...

work page

[8] [8]

The following ASR systems will be analyzed: • LF-MMI: This ASR refers to the traditional chain model using LF-MMI optimization criteria

Results In this section, the results of the end-to-end ASR are presented. The following ASR systems will be analyzed: • LF-MMI: This ASR refers to the traditional chain model using LF-MMI optimization criteria. The system is described in Section 4.4. This system is trained using the standard kaldi’s recipe [19]. • End-to-end: This refers to end-to-end ASR...

work page