Pith. sign in

REVIEW 4 major objections 7 minor 1 cited by

Universal-2-TF: Robust All-Neural Text Formatting for ASR

T0 review · 4 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read A two-stage all-neural pipeline—a shared-encoder classifier plus a span-level seq2seq converter—can handle punctuation, truecasing, and inverse text normalization more accurately and cheaply than rule-based hybrids.

desk verdict Solid industrial text-formatting paper with a real architectural contribution, but the full-seq2seq baseline is mistrained, so the headline efficiency and hallucination claims need a retrained baseline before publication. read the letter →

arxiv 2501.05948 v1 pith:7DSP72LM submitted 2025-01-10 cs.CL

classification cs.CL
keywords textformattinginversenormalizationpunctuationrestorationtruecasingautomaticspeechrecognitionsequence-to-sequencemulti-tasklearningspan-basedgeneration
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

This paper claims that text formatting for automatic speech recognition—adding punctuation, restoring correct capitalization, and converting spoken forms like "twelve point three million dollars" into "$12.3 million"—is best handled by a fully neural two-stage pipeline rather than by handcrafted rules or hybrid systems. The proposed Universal-2-TF uses a shared-encoder token classifier to label punctuation, casing, and inverse-text-normalization spans, then a sequence-to-sequence model that only rewrites the short flagged spans. Because generation is limited to those spans, the pipeline is meant to avoid the high cost and hallucination risk of whole-transcript seq2seq formatting while keeping the flexibility that rule-based finite-state transducers lack. If the claim holds, commercial ASR transcripts become more accurate, cheaper to produce, and cleaner to read.

What carries the argument

The load-bearing mechanism is the two-stage pipeline. The first stage is a Transformer encoder (initialized from BERT) with three lightweight classification heads: a post-punctuation head, a token-casing head, and an ITN-span head; these share one encoder so inference is cheaper than running separate models. The second stage is a BART-style seq2seq model with a bidirectional encoder and autoregressive decoder, applied only to short spans that the first stage flags for ITN or mixed-casing, using a small window of context. Restricting the seq2seq model to those spans is what keeps computational cost low and avoids hallucinated rewrites of long text, while the token-level casing head avoids expensive character-level decisions.

What would settle it

Take a set of transcripts from an actual speech recognizer in raw spoken form, have humans produce written-form references, run Universal-2-TF on them, and compare I-WER and M-WER against the numbers reported on corpora normalized with the same tool used during training; if the gap is much larger or the advantage over the previous hybrid system disappears, the training-distribution assumption fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that splitting text formatting into a cheap multi-objective labeling stage and a narrow seq2seq conversion stage gives the best of both worlds: the classifier handles punctuation, sentence capitalization, acronyms, and span detection with one shared encoder, while the generator handles only the difficult conversions—mixed-case words and inverse text normalization—on short spans with left and right context. The reported experiments show this two-stage design reduces ITN word error rate from 52.7% to 30.3% relative to the previous WFST-based system, improves punctuation error rate from 29.9% to 29.0%, and lowers inference time on short texts from 11.2 to 10.7 seconds and on long texts from 127.9 to 92.7 seconds. In a blind comparison, human judges preferred its output over the previous system in 81.2% of cases.

Load-bearing premise

The whole system assumes that the spoken-form text used for training and evaluation—produced by applying a particular text normalizer to written corpora—matches what real speech recognizers actually emit, so if real STT output differs systematically, the measured accuracy and speed gains may not appear in production.

Editorial extensions

If this is right

  • WFST-based inverse text normalization becomes replaceable by a learned span converter that generalizes to entity types such as credit card numbers, emails, and postal addresses without hand-authored grammars.
  • Whole-transcript sequence-to-sequence formatting looks unattractive for production: the same data and architecture applied end-to-end gave worse accuracy and roughly 20–30 times longer inference on long texts.
  • A shared multi-objective encoder can handle punctuation, casing, and span detection simultaneously with no accuracy loss compared to separate task models, lowering deployment cost.
  • Text-only data, cleaned and augmented with synthetic examples from language models, is sufficient to train the conversion stage, so TF capability can be expanded without paired audio.
  • Human-perceived transcript quality, not just token-level error rates, improves substantially when ITN and mixed-casing are handled by the span generator.

Reading between the lines

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

  • One consequence not explored in the paper is that the pipeline inherits a recall bottleneck: if the first-stage classifier misses an ITN or mixed-case span, the second stage can never correct it, so span-detection recall likely bounds overall formatting accuracy.
  • The span-gated design could generalize to other rewriting tasks in ASR, such as disfluency removal, speaker-label formatting, or non-English punctuation systems, by changing the classifier's label sets and retraining the span converter.
  • Because training and evaluation both rely on normalizer-generated spoken-form text, a direct test on raw outputs from a variety of speech recognizers would reveal how much of the reported advantage transfers to real production audio.
  • The I-WER of 30.3% still means roughly one in three ITN words is wrong on the public test sets, so reporting error rates per entity class would make the model's remaining weaknesses actionable rather than averaged away.
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.

Referee Report

4 major / 7 minor

Summary. The paper proposes Universal-2-TF, a two-stage all-neural text formatting pipeline for ASR. The first stage is a BERT-based multi-objective token classifier that jointly predicts punctuation, truecasing labels, and ITN span boundaries. The second stage is a BART-based seq2seq model that converts only the detected short spans to perform ITN and mixed-case restoration. The authors train on a large text corpus (10.2B words) generated by applying the NeMo text normalizer to written text, plus LLM-synthetic data. They evaluate the system on five public datasets and several internal ITN-focused sets, comparing against a full-seq2seq baseline, the previous Universal-1-TF system, and open-source component models. They report that Universal-2-TF achieves lower PER, CER, M-WER, and I-WER than the two internal baselines while being substantially faster, and that human judges prefer it over Universal-1-TF.

Significance. If the empirical claims are valid, this is a valuable applied contribution: it demonstrates a practical two-stage design that avoids full-transcript seq2seq cost while retaining neural flexibility for ITN, and it provides a holistic evaluation of punctuation, truecasing, and ITN together. The paper includes a useful ablation of shared parameterization in the first stage, transparent descriptions of the data processing pipeline, and a human preference study. However, the headline comparison to a full-seq2seq baseline is undermined by a training/evaluation mismatch, and the lack of statistical significance testing weakens the quantitative claims. The architecture itself is not entirely new, building on earlier span-based ITN work, but the joint treatment of PR, truecasing, ITN, and mixed-case in a commercial ASR pipeline is of practical interest.

major comments (4)
  1. [Section 5.1, Table 2] The 'full seq2seq' baseline is not a valid end-to-end comparison. The paper states that this model has an architecture identical to the second-stage seq2seq model and was trained on the same data used for training the Universal-2-TF seq2seq model, which is the span-level ITN/mixed-case dataset described in Section 4, not a dataset of full transcripts with punctuation and truecasing targets. Evaluating this span-trained model on entire normalized transcripts with punctuation and truecasing requirements conflates the two-stage architecture's benefit with a training/serving mismatch. The large degradations in PER, CER, M-WER, and I-WER, and the long inference times, are plausibly due to length and domain mismatch rather than an inherent weakness of end-to-end seq2seq TF. The claim that 'applying a seq2seq model in an end-to-end fashion ... results in degraded TF performance' is therefore not supported. Please retrain or otherwise obtain a fair full-transcript seq2seq baseline, or revise the claim accordingly.
  2. [Section 5.1, I-WER definition and evaluation inputs] The I-WER metric definition and the test data preparation are underspecified, which is load-bearing for the external validity of all reported numbers. Specifically, the paper does not state how the public written-text datasets (SummScreen, DialogSum, AMI, MeetingBank, Europarl) were converted into the 'normalized' inputs fed to the model. If the same NeMo text normalizer used to generate training data was also used for all test inputs, the evaluation is partly circular: the model is tuned to invert a particular normalization scheme, and accuracy on these datasets may not transfer to real STT outputs that are not produced by NeMo. The paper should specify the exact normalization procedure for each test set, and ideally include at least one test set derived from actual ASR output or independently transcribed speech. Additionally, no confidence intervals or significance tests are reported for any of the Table 2 numbers, so it is unclear whether the differences between Universal-2-TF and Universal-1-TF are meaningful.
  3. [Section 5.2, Table 3] The subjective evaluation lacks essential details for reproducibility and interpretive weight. The paper reports only the aggregate win rates (81.2% vs 17.2% with 1.6% neutral) and states that 400 samples were labeled by six judges each, with recruitment by two external vendors. It does not report inter-annotator agreement, the instruction protocol, the entity distribution of the 400 samples, or any statistical test on the preference counts. More importantly, the comparison is only against Universal-1-TF, so the perceptual quality advantages over a fairly trained full-seq2seq system remain untested. Given the central claim about 'superior perceptual quality', this evidence is currently thin.
  4. [Section 4.2 and Section 5.3] The contribution of the LLM-generated synthetic data is asserted but never isolated. The paper states that the seq2seq model was fine-tuned in two steps, the second of which used 'an ITN-focused dataset consisting of 2B words, specifically generated through simulation for ITN and mixed-case word conversion', and that synthetic data from multiple LLMs provided the model with an ability to handle practically important formatting cases. However, no ablation is presented (e.g., training without synthetic data, or with only the non-synthetic subset). Consequently, the claim that the LLM synthetic data is beneficial is not empirically established in the present paper. Since this is an important component of the method, an ablation or at least a sensitivity analysis should be reported.
minor comments (7)
  1. [Section 1, example] The illustrative example appears to be inconsistent: the input mentions 'aicorps's revenue reached twelve point three million dollars' while the output says 'AICorp's Q1 revenue reached $12.3 million', where 'Q1' is not present in the input. Please make the example self-consistent.
  2. [Figure 1] The extracted span example shows the word 'announced' duplicated in the input to the seq2seq model ('... Sarah mcallister announced announced twelve point three ...'). This appears to be a typo in the figure, which is confusing because span extraction should not duplicate tokens.
  3. [Section 5.1] The paper does not discuss the possibility of train/test overlap between the large public/purchased training corpora (e.g., CorpusData, Wikipedia) and the five public evaluation datasets. Since some of these datasets (e.g., SummScreen, MeetingBank) could conceivably be present in web-scale corpora, the authors should state whether any de-duplication of test sets against training data was performed.
  4. [Table 1] The checkmarks for 'Used for 2nd stage training' are not aligned with the column header, which makes the table hard to read. Please reformat the table for clarity.
  5. [Section 5.3, Table 6] The six private ITN test sets are said to contain 2,000 samples each, but no details are given on how they were constructed, what the distribution of entity types is, or whether they are available to other researchers. At least a brief description of the collection or simulation procedure would be useful for interpreting the results.
  6. [Section 5.3, paragraph after Table 4] The phrase 'avoiding penalties from the domain mismatch' is a bit unclear; consider rephrasing to 'avoiding the performance penalty caused by domain mismatch' for readability.
  7. [References] Some references contain line-break artifacts (e.g., 'Piotr ˙Zelasko ...' and 'Piotr P˛ezik') and the AssemblyAI 2024 reference lacks a URL or venue; please clean up the reference list.

Circularity Check

1 steps flagged · score 4.0 of 10

Full-seq2seq baseline is trained on span-level data but evaluated on full transcripts, making the claimed advantage over end-to-end seq2seq partly an artifact of the comparison protocol; the core TF results on public benchmarks remain independent.

  1. other [Section 5.1 (Objective Evaluation Results), paragraph 3; cf. Section 3.2 and Section 4]
    "The first model, full seq2seq, is a seq2seq model based on a Transformer encoder/decoder, which directly processes normalized input text and generates formatted text. Its architecture is identical to that of the second-stage seq2seq model in Universal-2-TF and was trained on the same data used for training the Universal-2-TF seq2seq model. The result indicates that applying a seq2seq model in an end-to-end fashion not only leads to significant inefficiency in inference but also results in degraded TF performance due to lack of robustness and hallucination."

    The 'same data' used to train the Universal-2-TF seq2seq model is span-level data: Section 3.2 defines the seq2seq input as 'a sequence of tokens in each span identified from the first stage,' and Table 1 lists second-stage training sets as full corpora that are converted into span input-output pairs. The full-seq2seq baseline is therefore optimized to convert short ITN/mixed-case spans, but it is evaluated on whole normalized transcripts of up to 5,478 words. The reported degradations (PER 35.0 vs 29.0, CER 2.5 vs 0.9, M-WER 2.3 vs 0.4, I-WER 37.6 vs 30.3, and runtime 222.9s/2845.8s vs 10.7s/92.7s) are a forced consequence of this train/evaluation distribution mismatch, not an inherent property of end-to-end seq2seq architectures.

full rationale

No circularity is found in the core derivation of Universal-2-TF itself: the two-stage model is trained on NeMo-normalized text and evaluated on external public datasets with written-form references, so its absolute accuracy numbers are independent evidence. The use of NeMo to generate both training inputs and evaluation inputs is a data-generation choice, not a tautology, and the paper does not claim to derive the NeMo normalizer from first principles. The main circularity-adjacent problem is the full-seq2seq comparison: the baseline is trained on the same span-level data as the proposed second-stage model, yet is tested on full transcripts, so the observed degradation is guaranteed by the mismatch in sequence-length and task distribution. This makes one load-bearing comparative claim partly an artifact of construction, while leaving the proposed model's performance on public benchmarks intact. No load-bearing self-citation chain or uniqueness argument is present; self-citations to prior AssemblyAI work are descriptive baseline references rather than mathematical premises. Score 4 reflects one by-construction comparison confound with otherwise independent central results.

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

The paper introduces no new physical or conceptual entities. Its central claims rest on the representativeness of NeMo-based data simulation, the accuracy of first-stage span detection, and the transferability of pretrained models, none of which are independently verified.

free parameters (3)
  • Task weights alpha_1, alpha_2, alpha_3 = 1/3 each
    Set to equal values for simplicity in Equation (1); not optimized, but affects the balance of the multi-task loss.
  • Span context size = one word left/right shown in Figure 1
    The seq2seq model receives limited left and right context around detected spans; the exact size is a design choice affecting conversion accuracy and cost.
  • Data filtering thresholds = not specified
    The cleaning pipeline uses heuristic thresholds for punctuation and capitalization frequency, developed by examining data samples, but the exact cutoffs are not given.
assumptions (5)
  • domain assumption NeMo Text Normalizer produces valid and representative spoken-form text for the target domains
    Section 4 uses NeMo to generate training pairs; Section 5 evaluates on datasets that are presumably normalized the same way. If real STT output differs, performance may degrade.
  • domain assumption The first-stage classifier reliably identifies all spans needing ITN or mixed-case conversion
    The second stage only sees spans selected by the first stage. Missed spans are never converted, and no analysis of span detection recall is provided.
  • domain assumption Pretrained BERT and BART weights transfer to the text formatting tasks
    The first stage is initialized with bert-base-uncased and the second with bart-base, assuming the pretrained representations are useful starting points.
  • ad hoc to paper LLM-generated synthetic data is high quality and useful for seq2seq training
    Section 4.2 adds synthetic examples from LLMs, but provides no analysis of the quality, accuracy, or distribution shift introduced by this data.
  • domain assumption Token-level MIXED labels plus seq2seq conversion fully solve truecasing for mixed-case words
    If a mixed-case word is assigned CAPITAL or LOWER by the first stage, the seq2seq model is never invoked, so the word will not receive its correct mixed-case form.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Universal-2-TF: Robust All-Neural Text Formatting for ASR." pith.science (2026). https://pith.science/paper/7DSP72LM

@misc{pith2026250105948,
  author       = {Pith},
  title        = {Pith review of: Universal-2-TF: Robust All-Neural Text Formatting for ASR},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7DSP72LM}},
  note         = {Machine review of arXiv:2501.05948}
}
read the original abstract

This paper introduces an all-neural text formatting (TF) model designed for commercial automatic speech recognition (ASR) systems, encompassing punctuation restoration (PR), truecasing, and inverse text normalization (ITN). Unlike traditional rule-based or hybrid approaches, this method leverages a two-stage neural architecture comprising a multi-objective token classifier and a sequence-to-sequence (seq2seq) model. This design minimizes computational costs and reduces hallucinations while ensuring flexibility and robustness across diverse linguistic entities and text domains. Developed as part of the Universal-2 ASR system, the proposed method demonstrates superior performance in TF accuracy, computational efficiency, and perceptual quality, as validated through comprehensive evaluations using both objective and subjective methods. This work underscores the importance of holistic TF models in enhancing ASR usability in practical settings.

Figures

Figures reproduced from arXiv: 2501.05948 by the authors.

Figure 1
Figure 1. Universal-2-TF model architecture: A Transformer-based encoder generates token representations of the [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Text formatting examples comparing Universal-2-TF (proposed model) with Universal-1-TF (previous [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Balalaika: Data-Centric, Prosody-Aware Annotation Pipeline for Russian Speech

    cs.CL 2025-07 unverdicted novelty 6.0 of 10

    Balalaika is a data-centric annotation pipeline for Russian speech that combines semantic VAD, ASR ensembling, and prosody enrichment to build a 5.1k-hour corpus showing gains in denoising and TTS.

Reference graph

Works this paper leans on

34 extracted references · 16 canonical work pages · cited by 1 Pith paper

  1. [1]

    Alexandra Antonova, Evelina Bakhturina, and Boris Ginsburg. 2022. https://doi.org/10.21437/INTERSPEECH.2022-10864 Thutmose tagger: Single-pass neural model for inverse text normalization . In 23rd Annual Conference of the International Speech Communication Association, Interspeech 2022, Incheon, Korea, September 18-22, 2022, pages 550--554. ISCA

  2. [2]

    AssemblyAI. 2024. https://www.assemblyai.com/blog/introducing-our-new-punctuation-restoration-and-truecasing-models/ Introducing our new punctuation restoration and truecasing models

  3. [3]

    Jean Carletta, Simone Ashby, Sebastien Bourban, Mike Flynn, Mael Guillemot, Thomas Hain, Jaroslav Kadlec, Vasilis Karaiskos, Wessel Kraaij, Melissa Kronenthal, Guillaume Lathoud, Mike Lincoln, Agnes Lisowska, Iain McCowan, Wilfried Post, Dennis Reidsma, and Pierre Wellner. 2006. The ami meeting corpus: A pre-announcement. In Machine Learning for Multimoda...

  4. [4]

    Mingda Chen, Zewei Chu, Sam Wiseman, and Kevin Gimpel. 2022. https://doi.org/10.18653/v1/2022.acl-long.589 S umm S creen: A dataset for abstractive screenplay summarization . In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8602--8615, Dublin, Ireland. Association for Computational L...

  5. [5]

    Yulong Chen, Yang Liu, Liang Chen, and Yue Zhang. 2021. https://doi.org/10.18653/v1/2021.findings-acl.449 D ialog S um: A real-life scenario dialogue summarization dataset . In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 5062--5074, Online. Association for Computational Linguistics

  6. [6]

    Maury Courtland, Adam Faulkner, and Gayle McElvain. 2020. https://doi.org/10.18653/v1/2020.iwslt-1.33 Efficient automatic punctuation restoration using bidirectional transformers with robust inference . In Proceedings of the 17th International Conference on Spoken Language Translation, pages 272--279, Online. Association for Computational Linguistics

  7. [7]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. https://doi.org/10.18653/v1/N19-1423 BERT : Pre-training of deep bidirectional transformers for language understanding . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long a...

  8. [9]

    Nuno Miguel Guerreiro, Ricardo Rei, and Fernando Batista. 2021. https://doi.org/https://doi.org/10.1016/j.eswa.2021.115740 Towards better subtitles: A multilingual approach for punctuation restoration of speech transcripts . Expert Systems with Applications, 186:115740

Show all 34 references
  1. [10]

    Oliver Guhr, Anne-Kathrin Schumann, Frank Bahrmann, and Hans Joachim Böhme. 2021. http://ceur-ws.org/Vol-2957/sepp_paper4.pdf Fullstop: Multilingual deep models for punctuation prediction . In Proceedings of the Swiss Text Analytics Conference 2021, Winterthur, Switzerland. CE...

  2. [11]

    Yebowen Hu, Timothy Ganter, Hanieh Deilamsalehy, Franck Dernoncourt, Hassan Foroosh, and Fei Liu. 2023. https://doi.org/10.18653/v1/2023.acl-long.906 M eeting B ank: A benchmark dataset for meeting summarization . In Proceedings of the 61st Annual Meeting of the Association fo...

  3. [12]

    Philipp Koehn. 2005. https://aclanthology.org/2005.mtsummit-papers.11/ E uroparl: A parallel corpus for statistical machine translation . In Proceedings of Machine Translation Summit X: Papers, pages 79--86, Phuket, Thailand

  4. [13]

    Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, and Luke Zettlemoyer. 2020. https://doi.org/10.18653/v1/2020.acl-main.703 BART : Denoising sequence-to-sequence pre-training for natural language generation, translatio...

  5. [14]

    Binghuai Lin and Liyuan Wang. 2020. https://doi.org/10.21437/Interspeech.2020-1277 Joint prediction of punctuation and disfluency in speech transcripts . In Interspeech 2020, pages 716--720

  6. [15]

    Aleksandr Meister, Matvei Novikov, Nikolay Karpov, Evelina Bakhturina, Vitaly Lavrukhin, and Boris Ginsburg. 2023. https://arxiv.org/abs/2310.02943 Librispeech-pc: Benchmark for evaluation of punctuation and capitalization capabilities of end-to-end asr models . arXiv preprint...

  7. [16]

    Binh Nguyen, Vu Bao Hung Nguyen, Hien Nguyen, Pham Ngoc Phuong, The-Loc Nguyen, Quoc Truong Do, and Luong Chi Mai. 2019. https://doi.org/10.1109/O-COCOSDA46868.2019.9041202 Fast and accurate capitalization and punctuation for automatic speech recognition using transformer and ...

  8. [17]

    Thai-Binh Nguyen, Quang Minh Nguyen, Quoc Truong Do, Chi Mai Luong, Alexander Waibel, et al. 2023. https://openreview.net/forum?id=VI9IWawOr3 Adapitn: A fast, reliable, and dynamic adaptive inverse text normalization . In ICASSP 2023-2023 IEEE International Conference on Acous...

  9. [18]

    Raghavendra Pappagari, Piotr Żelasko, Agnieszka Mikołajczyk, Piotr Pęzik, and Najim Dehak. 2021. https://doi.org/10.1109/ASRU51503.2021.9687976 Joint prediction of truecasing and punctuation for conversational speech in low-resource scenarios . In 2021 IEEE Automatic Speech Re...

  10. [19]

    Debjyoti Paul, Yutong Pang, Szu-Jui Chen, and Xuedong Zhang. 2022. https://www.isca-archive.org/interspeech_2022/paul22_interspeech.html Improving data driven inverse text normalization using data augmentation and machine translation . In Interspeech 2022, pages 5221--5222

  11. [20]

    Ernest Pusateri, Bharat Ram Ambati, Elizabeth Brooks, Ondrej Platek, Donald McAllaster, and Venki Nagesha. 2017. https://doi.org/10.21437/Interspeech.2017-1274 A mostly data-driven approach to inverse text normalization . In Interspeech 2017, pages 2784--2788

  12. [21]

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2023. https://dl.acm.org/doi/10.5555/3618408.3619590 Robust speech recognition via large-scale weak supervision . In Proceedings of the 40th International Conference on Machine Learning...

  13. [22]

    Gopi Ramena, Divija Nagaraju, Sukumar Moharana, Debi Prasanna Mohanty, and Naresh Purre. 2020. https://doi.org/10.1109/ICSC.2020.00035 An Efficient Architecture for Predicting the Case of Characters using Sequence Models . In 2020 IEEE 14th International Conference on Semantic...

  14. [23]

    Francis McCann Ramirez, Luka Chkhetiani, Andrew Ehrenberg, Robert McHardy, Rami Botros, Yash Khare, Andrea Vanzo, Taufiquzzaman Peyash, Gabriel Oexle, Michael Liang, Ilya Sklyar, Enver Fakhan, Ahmed Etefy, Daniel McCrystal, Sam Flamini, Domenic Donato, and Takuya Yoshioka. 202...

  15. [24]

    Richard Sproat and Navdeep Jaitly. 2017. http://arxiv.org/abs/1611.00068 Rnn approaches to text normalization: A challenge . arXiv preprint arXiv:1611.00068

  16. [25]

    Monica Sunkara, Srikanth Ronanki, Kalpit Dixit, Sravan Bodapati, and Katrin Kirchhoff. 2020. https://doi.org/10.18653/v1/2020.nlpmc-1.8 Robust prediction of punctuation and truecasing for medical ASR . In Proceedings of the First Workshop on Natural Language Processing for Med...

  17. [26]

    Monica Sunkara, Chaitanya Shivade, Sravan Bodapati, and Katrin Kirchhoff. 2021. https://doi.org/10.1109/ICASSP39728.2021.9414912 Neural inverse text normalization . In ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 7...

  18. [27]

    Raymond Hendy Susanto, Hai Leong Chieu, and Wei Lu. 2016. https://doi.org/10.18653/v1/D16-1225 Learning to capitalize with character-level recurrent neural networks: An empirical study . In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing,...

  19. [28]

    Sharman Tan, Piyush Behre, Nick Kibre, Issac Alphonso, and Shuangyu Chang. 2023. https://doi.org/10.1109/SLT54892.2023.10023257 Four-in-one: a joint approach to inverse text normalization, punctuation, capitalization, and disfluency for automatic speech recognition . In 2022 I...

  20. [29]

    Ottokar Tilk and Tanel Alumäe. 2015. https://doi.org/10.21437/Interspeech.2015-240 Lstm for punctuation restoration in speech transcripts . In Interspeech 2015, pages 683--687

  21. [30]

    Ronny Huang, Mingqing Chen, and Rajiv Mathews

    Hao Zhang, You-Chi Cheng, Shankar Kumar, W. Ronny Huang, Mingqing Chen, and Rajiv Mathews. 2022. https://doi.org/10.1109/ICASSP43922.2022.9746492 Capitalization normalization for language modeling with an accurate and efficient hierarchical rnn model . In ICASSP 2022 - 2022 IE...

  22. [31]

    Yang Zhang, Evelina Bakhturina, Kyle Gorman, and Boris Ginsburg. 2021. http://arxiv.org/abs/2104.05055 Nemo inverse text normalization: From development to production . CoRR, abs/2104.05055

  23. [32]

    Xiliang Zhu, Chia-Tien Chang, Shayna Gardiner, David Rossouw, and Jonas Robertson. 2024. https://aclanthology.org/2024.unimplicit-1.3/ Resolving transcription ambiguity in S panish: A hybrid acoustic-lexical system for punctuation restoration . In Proceedings of the Third Work...

  24. [33]

    Piotr Żelasko, Piotr Szymański, Jan Mizgajski, Adrian Szymczak, Yishay Carmiel, and Najim Dehak. 2018. https://doi.org/10.21437/Interspeech.2018-1096 Punctuation prediction model for conversational speech . In Interspeech 2018, pages 2633--2637

  25. [34]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before...

  26. [35]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.