From 3b3ada765cfbc523023e46465e88b449ea52ccba Mon Sep 17 00:00:00 2001 From: marcoyang Date: Wed, 28 Jun 2023 17:25:42 +0800 Subject: [PATCH] minor fixes --- .../decoding-with-langugage-models/LODR.rst | 2 +- .../shallow-fusion.rst | 10 +++++----- .../librispeech/distillation.rst | 6 +++--- .../pruned_transducer_stateless.rst | 18 +++++++++--------- .../recipes/Streaming-ASR/introduction.rst | 4 ++-- .../pruned_transducer_stateless.rst | 10 +++++----- .../librispeech/zipformer_transducer.rst | 4 ++-- 7 files changed, 27 insertions(+), 27 deletions(-) diff --git a/docs/source/decoding-with-langugage-models/LODR.rst b/docs/source/decoding-with-langugage-models/LODR.rst index 9e853bc24..4708963cb 100644 --- a/docs/source/decoding-with-langugage-models/LODR.rst +++ b/docs/source/decoding-with-langugage-models/LODR.rst @@ -45,7 +45,7 @@ Here, the source domain LM is trained on the training corpus. The only differenc shallow fusion is the subtraction of the source domain LM. Some works treat the predictor and the joiner of the neural transducer as its internal LM. However, the LM is -considered to be weak and can only capture low-level language information. Therefore, `LODR `_ proposed to use +considered to be weak and can only capture low-level language information. Therefore, `LODR `__ proposed to use a low-order n-gram LM as an approximation of the ILM of the neural transducer. This leads to the following formula during decoding for RNNT model: diff --git a/docs/source/decoding-with-langugage-models/shallow-fusion.rst b/docs/source/decoding-with-langugage-models/shallow-fusion.rst index 2a7a7e72c..8b226efc0 100644 --- a/docs/source/decoding-with-langugage-models/shallow-fusion.rst +++ b/docs/source/decoding-with-langugage-models/shallow-fusion.rst @@ -24,7 +24,7 @@ to improve the word-error-rate of a RNN Transducer model. We recommend you to use a GPU for decoding. -For illustration purpose, we will use a pre-trained ASR model from this `link `_. +For illustration purpose, we will use a pre-trained ASR model from this `link `__. If you want to train your model from scratch, please have a look at :ref:`non_streaming_librispeech_pruned_transducer_stateless`. As the initial step, let's download the pre-trained model. @@ -59,7 +59,7 @@ The following WERs are achieved on test-clean and test-other: $ beam_size_4 7.93 best for test-other These are already good numbers! But we can further improve it by using shallow fusion with external LM. -Training a language model usually takes a long time, we can download a pre-trained LM from this `link `_. +Training a language model usually takes a long time, we can download a pre-trained LM from this `link `__. .. code-block:: bash @@ -72,8 +72,8 @@ Training a language model usually takes a long time, we can download a pre-train .. note:: This is an RNN LM trained on the LibriSpeech text corpus. So it might not be ideal for other corpus. - You may also train a RNN LM from scratch. Please refer to this `script `_ - for training a RNN LM and this `script `_ to train a transformer LM. + You may also train a RNN LM from scratch. Please refer to this `script `__ + for training a RNN LM and this `script `__ to train a transformer LM. To use shallow fusion for decoding, we can execute the following command: @@ -141,7 +141,7 @@ A few parameters can be tuned to further boost the performance of shallow fusion Here, we also show how `--beam-size` effect the WER and decoding time: .. list-table:: WERs and decoding time (on test-clean) of shallow fusion with different beam sizes - :widths: 25 25 50 + :widths: 25 25 25 25 :header-rows: 1 * - Beam size diff --git a/docs/source/recipes/Non-streaming-ASR/librispeech/distillation.rst b/docs/source/recipes/Non-streaming-ASR/librispeech/distillation.rst index ca16341e9..2e8d0893a 100644 --- a/docs/source/recipes/Non-streaming-ASR/librispeech/distillation.rst +++ b/docs/source/recipes/Non-streaming-ASR/librispeech/distillation.rst @@ -13,7 +13,7 @@ for more details about MVQ-KD. `pruned_transducer_stateless4 `_. Currently, we only implement MVQ-KD in this recipe. However, MVQ-KD is theoretically applicable to all recipes with only minor changes needed. Feel free to try out MVQ-KD in different recipes. If you - encounter any problems, please open an issue here `icefall `_. + encounter any problems, please open an issue here `icefall `__. .. note:: @@ -217,7 +217,7 @@ the following command. --exp-dir $exp_dir \ --enable-distillation True -You should get similar results as `here `_. +You should get similar results as `here `__. That's all! Feel free to experiment with your own setups and report your results. -If you encounter any problems during training, please open up an issue `here `_. +If you encounter any problems during training, please open up an issue `here `__. diff --git a/docs/source/recipes/Non-streaming-ASR/librispeech/pruned_transducer_stateless.rst b/docs/source/recipes/Non-streaming-ASR/librispeech/pruned_transducer_stateless.rst index 42fd3df77..1bc1dd984 100644 --- a/docs/source/recipes/Non-streaming-ASR/librispeech/pruned_transducer_stateless.rst +++ b/docs/source/recipes/Non-streaming-ASR/librispeech/pruned_transducer_stateless.rst @@ -8,10 +8,10 @@ with the `LibriSpeech `_ dataset. .. Note:: - The tutorial is suitable for `pruned_transducer_stateless `_, - `pruned_transducer_stateless2 `_, - `pruned_transducer_stateless4 `_, - `pruned_transducer_stateless5 `_, + The tutorial is suitable for `pruned_transducer_stateless `__, + `pruned_transducer_stateless2 `__, + `pruned_transducer_stateless4 `__, + `pruned_transducer_stateless5 `__, We will take pruned_transducer_stateless4 as an example in this tutorial. .. HINT:: @@ -237,7 +237,7 @@ them, please modify ``./pruned_transducer_stateless4/train.py`` directly. .. NOTE:: - The options for `pruned_transducer_stateless5 `_ are a little different from + The options for `pruned_transducer_stateless5 `__ are a little different from other recipes. It allows you to configure ``--num-encoder-layers``, ``--dim-feedforward``, ``--nhead``, ``--encoder-dim``, ``--decoder-dim``, ``--joiner-dim`` from commandline, so that you can train models with different size with pruned_transducer_stateless5. @@ -529,13 +529,13 @@ Download pretrained models If you don't want to train from scratch, you can download the pretrained models by visiting the following links: - - `pruned_transducer_stateless `_ + - `pruned_transducer_stateless `__ - - `pruned_transducer_stateless2 `_ + - `pruned_transducer_stateless2 `__ - - `pruned_transducer_stateless4 `_ + - `pruned_transducer_stateless4 `__ - - `pruned_transducer_stateless5 `_ + - `pruned_transducer_stateless5 `__ See ``_ for the details of the above pretrained models diff --git a/docs/source/recipes/Streaming-ASR/introduction.rst b/docs/source/recipes/Streaming-ASR/introduction.rst index e1382e77d..ac77a51d1 100644 --- a/docs/source/recipes/Streaming-ASR/introduction.rst +++ b/docs/source/recipes/Streaming-ASR/introduction.rst @@ -45,9 +45,9 @@ the input features. We have three variants of Emformer models in ``icefall``. - - ``pruned_stateless_emformer_rnnt2`` using Emformer from torchaudio, see `LibriSpeech recipe `_. + - ``pruned_stateless_emformer_rnnt2`` using Emformer from torchaudio, see `LibriSpeech recipe `__. - ``conv_emformer_transducer_stateless`` using ConvEmformer implemented by ourself. Different from the Emformer in torchaudio, ConvEmformer has a convolution in each layer and uses the mechanisms in our reworked conformer model. - See `LibriSpeech recipe `_. + See `LibriSpeech recipe `__. - ``conv_emformer_transducer_stateless2`` using ConvEmformer implemented by ourself. The only difference from the above one is that it uses a simplified memory bank. See `LibriSpeech recipe `_. diff --git a/docs/source/recipes/Streaming-ASR/librispeech/pruned_transducer_stateless.rst b/docs/source/recipes/Streaming-ASR/librispeech/pruned_transducer_stateless.rst index de7102ba8..2ca70bcf3 100644 --- a/docs/source/recipes/Streaming-ASR/librispeech/pruned_transducer_stateless.rst +++ b/docs/source/recipes/Streaming-ASR/librispeech/pruned_transducer_stateless.rst @@ -6,10 +6,10 @@ with the `LibriSpeech `_ dataset. .. Note:: - The tutorial is suitable for `pruned_transducer_stateless `_, - `pruned_transducer_stateless2 `_, - `pruned_transducer_stateless4 `_, - `pruned_transducer_stateless5 `_, + The tutorial is suitable for `pruned_transducer_stateless `__, + `pruned_transducer_stateless2 `__, + `pruned_transducer_stateless4 `__, + `pruned_transducer_stateless5 `__, We will take pruned_transducer_stateless4 as an example in this tutorial. .. HINT:: @@ -264,7 +264,7 @@ them, please modify ``./pruned_transducer_stateless4/train.py`` directly. .. NOTE:: - The options for `pruned_transducer_stateless5 `_ are a little different from + The options for `pruned_transducer_stateless5 `__ are a little different from other recipes. It allows you to configure ``--num-encoder-layers``, ``--dim-feedforward``, ``--nhead``, ``--encoder-dim``, ``--decoder-dim``, ``--joiner-dim`` from commandline, so that you can train models with different size with pruned_transducer_stateless5. diff --git a/docs/source/recipes/Streaming-ASR/librispeech/zipformer_transducer.rst b/docs/source/recipes/Streaming-ASR/librispeech/zipformer_transducer.rst index f0e8961d7..8b75473c6 100644 --- a/docs/source/recipes/Streaming-ASR/librispeech/zipformer_transducer.rst +++ b/docs/source/recipes/Streaming-ASR/librispeech/zipformer_transducer.rst @@ -6,7 +6,7 @@ with the `LibriSpeech `_ dataset. .. Note:: - The tutorial is suitable for `pruned_transducer_stateless7_streaming `_, + The tutorial is suitable for `pruned_transducer_stateless7_streaming `__, .. HINT:: @@ -642,7 +642,7 @@ Download pretrained models If you don't want to train from scratch, you can download the pretrained models by visiting the following links: - - `pruned_transducer_stateless7_streaming `_ + - `pruned_transducer_stateless7_streaming `__ See ``_ for the details of the above pretrained models