Automatic Correction of Speech Recognized Mathematical Equations using Encoder-Decoder Attention Model

Mounika, Y.; Tarakaram, Y.; Prasanna, Y. Lakshmi; Gupta, Deepa; Pati, Peeta Basa

doi:10.1109/indicon56171.2022.10039926

Cited by 8 publications

(2 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A precision of 71.57%, recall of 38.65%, f1-score of 61.15% with respect to CoNLL-2014 and GLEU score of 61.00 with respect to JFLEG was achieved with such a combination. Mounika et al [25] have experimented with both pre-trained and fine-tuned T5 models for lowering the word error rate (WER) [26] in text generated by recognition of speech samples of mathematical equations. On both the models, GloVe & FastText embeddings were used to increase the accuracy.…”

Section: Tan Et Al'smentioning

confidence: 99%

“…These error types are: Typographic error, Cognitive error, Visual error, Runon error, Split-word error, Non-word error, Real-word error. Subsequently, Mounika et al [25] studied the error types present in automatically speech recognized (ASR) text in their work. Their work provided a list of errors observed in sentences.…”

Section: Error Category Definition and Shift Analysismentioning

confidence: 99%

See 1 more Smart Citation

Grammatical versus Spelling Error Correction: An Investigation into the Responsiveness of Transformer-Based Language Models Using BART and MarianMT

Raju,

Pati,

Gandheesh

et al. 2024

J. Info. Know. Mgmt.

View full text Add to dashboard Cite

Text continues to remain a relevant form of representation for information. Text documents are created either in digital native platforms or through the conversion of other media files such as images and speech. While the digital native text is invariably obtained through physical or virtual keyboards, technologies such as OCR and speech recognition are utilised to transform the images and speech signals into text content. All these variety of mechanisms of text generation also introduce errors into the captured text. This project aims at analysing different kinds of errors that occur in text documents. The work employs two of the advanced deep neural network-based language models, namely, BART and MarianMT, to rectify the anomalies present in the text. Transfer learning of these models with available dataset is performed to finetune their capacity for error correction. A comparative study is conducted to investigate the effectiveness of these models in handling each of the defined error categories. It is observed that while both models can bring down the erroneous sentences by 20+%, BART can handle spelling errors far better (24.6%) than grammatical errors (8.8%).

show abstract

Section: Tan Et Al'smentioning

confidence: 99%

Section: Error Category Definition and Shift Analysismentioning

confidence: 99%