mirror of
https://github.com/storytold/storyteller-ml.git
synced 2026-10-13 23:53:02 +00:00
Introduce code from Rust monorepo. This is a mess
This commit is contained in:
Brandon Thomas
+35
@@ -0,0 +1,35 @@
|
||||
NeMo Text Processing API
|
||||
========================
|
||||
|
||||
Model Classes
|
||||
-------------
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.GraphFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.ClassifyFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.VerbalizeFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.VerbalizeFinalFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.inverse_text_normalization.ClassifyFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.inverse_text_normalization.VerbalizeFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.inverse_text_normalization.VerbalizeFinalFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
@@ -0,0 +1,17 @@
|
||||
Text Processing
|
||||
===============
|
||||
|
||||
`nemo_text_processing` is a python package that is installed with the `nemo_toolkit`.
|
||||
|
||||
See :doc:`NeMo Introduction <../starthere/intro>` for installation details.
|
||||
|
||||
Additional requirements can be found in `setup.sh <https://github.com/NVIDIA/NeMo/blob/main/nemo_text_processing/setup.sh>`_.
|
||||
|
||||
.. toctree::
|
||||
:maxdepth: 1
|
||||
|
||||
text_normalization
|
||||
inverse_text_normalization
|
||||
api
|
||||
|
||||
|
||||
@@ -0,0 +1,90 @@
|
||||
Inverse Text Normalization
|
||||
==========================
|
||||
|
||||
Inverse text normalization (ITN) is a part of the Automatic Speech Recognition (ASR) post-processing pipeline.
|
||||
ITN is the task of converting the raw spoken output of the ASR model into its written form to improve text readability.
|
||||
|
||||
For example,
|
||||
`"in nineteen seventy"` -> `"in 1975"`
|
||||
and `"it costs one hundred and twenty three dollars"` -> `"it costs $123"`.
|
||||
|
||||
NeMo ITN :cite:`textprocessing-itn-zhang2021nemo` is based on WFST-grammars :cite:`textprocessing-itn-Mohri2009`. We also provide a deployment route to C++ using `Sparrowhawk <https://github.com/google/sparrowhawk>`_ :cite:`textprocessing-itn-sparrowhawk` -- an open-source version of Google Kestrel :cite:`textprocessing-itn-ebden2015kestrel`.
|
||||
See :doc:`Text Procesing Deployment <../tools/text_processing_deployment>` for details.
|
||||
|
||||
.. note::
|
||||
|
||||
For more details, see the tutorial `NeMo/tutorials/text_processing/Inverse_Text_Normalization.ipynb <https://github.com/NVIDIA/NeMo/blob/main/tutorials/text_processing/Inverse_Text_Normalization.ipynb>`__ in `Google's Colab <https://colab.research.google.com/github/NVIDIA/NeMo/blob/main/tutorials/text_processing/Inverse_Text_Normalization.ipynb>`_.
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
Classes
|
||||
----------------------------------
|
||||
|
||||
|
||||
The base class for every grammar is :class:`GraphFst<nemo_text_processing.text_normalization.GraphFst>`.
|
||||
This tool is designed as a two-stage application: 1. `classification` of the input into semiotic tokens and 2. `verbalization` into written form.
|
||||
For every stage and every semiotic token class there is a corresponding grammar, e.g. :class:`taggers.CardinalFst<nemo_text_processing.inverse_text_normalization.taggers.cardinal.CardinalFst>`
|
||||
and :class:`verbalizers.CardinalFst<nemo_text_processing.inverse_text_normalization.verbalizers.cardinal.CardinalFst>`.
|
||||
Together, they compose the final grammars :class:`ClassifyFst<nemo_text_processing.inverse_text_normalization.ClassifyFst>` and
|
||||
:class:`VerbalizeFinalFst<nemo_text_processing.inverse_text_normalization.VerbalizeFinalFst>` that are compiled into WFST and used for inference.
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
.. autoclass:: nemo_text_processing.inverse_text_normalization.ClassifyFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.inverse_text_normalization.VerbalizeFinalFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
|
||||
Prediction
|
||||
----------------------------------
|
||||
|
||||
Example prediction run:
|
||||
|
||||
.. code::
|
||||
|
||||
python run_prediction.py --input=<INPUT_TEXT_FILE> --output=<OUTPUT_PATH> [--verbose]
|
||||
|
||||
|
||||
The input is expected to be lower-cased. Punctuation are outputted with separating spaces after semiotic tokens, e.g. `"i see, it is ten o'clock..."` -> `"I see, it is 10:00 . . ."`.
|
||||
Inner-sentence white-space characters in the input are not maintained.
|
||||
|
||||
|
||||
|
||||
Data Cleaning for Evaluation
|
||||
----------------------------------
|
||||
|
||||
.. code::
|
||||
|
||||
python clean_eval_data.py --input=<INPUT_TEXT_FILE>
|
||||
|
||||
|
||||
|
||||
Evaluation
|
||||
----------------------------------
|
||||
|
||||
Example evaluation run on (cleaned) `Google's text normalization dataset <https://www.kaggle.com/richardwilliamsproat/text-normalization-for-english-russian-and-polish>`__ :cite:`textprocessing-itn-sproat2016rnn`:
|
||||
|
||||
.. code::
|
||||
|
||||
python run_evaluation.py --input=./en_with_types/output-00001-of-00100 [--cat CLASS_CATEGORY] [--filter]
|
||||
|
||||
|
||||
|
||||
|
||||
References
|
||||
----------
|
||||
|
||||
.. bibliography:: textprocessing_all.bib
|
||||
:style: plain
|
||||
:labelprefix: TEXTPROCESSING-ITN
|
||||
:keyprefix: textprocessing-itn-
|
||||
@@ -0,0 +1,78 @@
|
||||
Text Normalization
|
||||
==================
|
||||
|
||||
NeMo Text Normalization converts text from written form into its verbalized form. It is used as a preprocessing step before Text to Speech (TTS). It could also be used for preprocessing Automatic Speech Recognition (ASR) training transcripts.
|
||||
|
||||
|
||||
For example,
|
||||
`"at 10:00"` -> `"at ten o'clock"`
|
||||
and `"it weighs 10kg."` -> `"it weights ten kilograms ."`.
|
||||
|
||||
|
||||
NeMo Text Normalization :cite:`textprocessing-norm-zhang2021nemo` is based on WFST-grammars :cite:`textprocessing-norm-Mohri2009`. We also provide a deployment route to C++ using `Sparrowhawk <https://github.com/google/sparrowhawk>`_ :cite:`textprocessing-norm-sparrowhawk` -- an open-source version of Google Kestrel :cite:`textprocessing-norm-ebden2015kestrel`.
|
||||
See :doc:`Text Procesing Deployment <../tools/text_processing_deployment>` for details.
|
||||
|
||||
|
||||
.. note::
|
||||
|
||||
For more details, see the tutorial `NeMo/tutorials/text_processing/Text_Normalization.ipynb <https://github.com/NVIDIA/NeMo/blob/main/tutorials/text_processing/Text_Normalization.ipynb>`__ in `Google's Colab <https://colab.research.google.com/github/NVIDIA/NeMo/blob/main/tutorials/text_processing/Text_Normalization.ipynb>`_.
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
Classes
|
||||
----------------------------------
|
||||
|
||||
|
||||
The base class for every grammar is :class:`GraphFst<nemo_text_processing.text_normalization.GraphFst>`.
|
||||
This tool is designed as a two-stage application: 1. `classification` of the input into semiotic tokens and 2. `verbalization` into written form.
|
||||
For every stage and every semiotic token class there is a corresponding grammar, e.g. :class:`taggers.CardinalFst<nemo_text_processing.text_normalization.taggers.cardinal.CardinalFst>`
|
||||
and :class:`verbalizers.CardinalFst<nemo_text_processing.text_normalization.verbalizers.cardinal.CardinalFst>`.
|
||||
Together, they compose the final grammars :class:`ClassifyFst<nemo_text_processing.text_normalization.ClassifyFst>` and
|
||||
:class:`VerbalizeFinalFst<nemo_text_processing.text_normalization.VerbalizeFinalFst>` that are compiled into WFST and used for inference.
|
||||
|
||||
|
||||
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.ClassifyFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
.. autoclass:: nemo_text_processing.text_normalization.VerbalizeFinalFst
|
||||
:show-inheritance:
|
||||
:members:
|
||||
|
||||
|
||||
Prediction
|
||||
----------------------------------
|
||||
|
||||
Example prediction run:
|
||||
|
||||
.. code::
|
||||
|
||||
python run_prediction.py <--input INPUT_TEXT_FILE> <--output OUTPUT_PATH> [--input_case INPUT_CASE]
|
||||
|
||||
``INPUT_CASE`` specifies whether to treat the input as lower-cased or case sensitive. By default treat the input as cased since this is more informative, especially for abbreviations. Punctuation are outputted with separating spaces after semiotic tokens, e.g. `"I see, it is 10:00..."` -> `"I see, it is ten o'clock . . ."`.
|
||||
Inner-sentence white-space characters in the input are not maintained.
|
||||
|
||||
|
||||
Evaluation
|
||||
----------------------------------
|
||||
|
||||
Example evaluation run on `Google's text normalization dataset <https://www.kaggle.com/richardwilliamsproat/text-normalization-for-english-russian-and-polish>`__ :cite:`textprocessing-norm-sproat2016rnn`:
|
||||
|
||||
.. code::
|
||||
|
||||
python run_evaluation.py --input=./en_with_types/output-00001-of-00100 [--cat CLASS_CATEGORY] [--input_case INPUT_CASE]
|
||||
|
||||
|
||||
|
||||
References
|
||||
----------
|
||||
|
||||
.. bibliography:: textprocessing_all.bib
|
||||
:style: plain
|
||||
:labelprefix: TEXTPROCESSING-NORM
|
||||
:keyprefix: textprocessing-norm-
|
||||
@@ -0,0 +1,58 @@
|
||||
@article{ebden2015kestrel,
|
||||
title={The Kestrel TTS text normalization system},
|
||||
author={Ebden, Peter and Sproat, Richard},
|
||||
journal={Natural Language Engineering},
|
||||
volume={21},
|
||||
number={3},
|
||||
pages={333},
|
||||
year={2015},
|
||||
publisher={Cambridge University Press}
|
||||
}
|
||||
|
||||
|
||||
@article{sproat2016rnn,
|
||||
title={RNN approaches to text normalization: A challenge},
|
||||
author={Sproat, Richard and Jaitly, Navdeep},
|
||||
journal={arXiv preprint arXiv:1611.00068},
|
||||
year={2016}
|
||||
}
|
||||
|
||||
@book{taylor2009text,
|
||||
title={Text-to-speech synthesis},
|
||||
author={Taylor, Paul},
|
||||
year={2009},
|
||||
publisher={Cambridge university press}
|
||||
}
|
||||
|
||||
@misc{zhang2021nemo,
|
||||
title={NeMo Inverse Text Normalization: From Development To Production},
|
||||
author={Yang Zhang and Evelina Bakhturina and Kyle Gorman and Boris Ginsburg},
|
||||
year={2021},
|
||||
eprint={2104.05055},
|
||||
archivePrefix={arXiv},
|
||||
primaryClass={cs.CL}
|
||||
}
|
||||
|
||||
@Inbook{Mohri2009,
|
||||
author="Mohri, Mehryar",
|
||||
editor="Droste, Manfred
|
||||
and Kuich, Werner
|
||||
and Vogler, Heiko",
|
||||
title="Weighted Automata Algorithms",
|
||||
bookTitle="Handbook of Weighted Automata",
|
||||
year="2009",
|
||||
publisher="Springer Berlin Heidelberg",
|
||||
address="Berlin, Heidelberg",
|
||||
pages="213--254",
|
||||
abstract="Weighted automata and transducers are widely used in modern applications in bioinformatics and text, speech, and image processing. This chapter describes several fundamental weighted automata and shortest-distance algorithms including composition, determinization, minimization, and synchronization, as well as single-source and all-pairs shortest distance algorithms over general semirings. It presents the pseudocode of these algorithms, gives an analysis of their running time complexity, and illustrates their use in some simple cases. Many other complex weighted automata and transducer algorithms used in practice can be obtained by combining these core algorithms.",
|
||||
isbn="978-3-642-01492-5",
|
||||
doi="10.1007/978-3-642-01492-5_6",
|
||||
url="https://doi.org/10.1007/978-3-642-01492-5_6"
|
||||
}
|
||||
|
||||
@inproceedings{sparrowhawk,
|
||||
title = {TTS for Low Resource Languages: A Bangla Synthesizer},
|
||||
author = {Alexander Gutkin and Linne Ha and Martin Jansche and Knot Pipatsrisawat and Richard Sproat},
|
||||
booktitle = {10th Language Resources and Evaluation Conference},
|
||||
year = {2016},
|
||||
}
|
||||
Reference in New Issue
Block a user