Opset17へのエクスポート、tiktokenへの置き換え、timestampルールの更新

ORIGINAL.md

@@ -0,0 +1,132 @@
+# Whisper Original information
+
+[[Blog]](https://openai.com/blog/whisper)
+[[Paper]](https://cdn.openai.com/papers/whisper.pdf)
+[[Model card]](model-card.md)
+[[Colab example]](https://colab.research.google.com/github/openai/whisper/blob/master/notebooks/LibriSpeech.ipynb)
+
+Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification.
+
+
+## Approach
+
+![Approach](approach.png)
+
+A Transformer sequence-to-sequence model is trained on various speech processing tasks, including multilingual speech recognition, speech translation, spoken language identification, and voice activity detection. All of these tasks are jointly represented as a sequence of tokens to be predicted by the decoder, allowing for a single model to replace many different stages of a traditional speech processing pipeline. The multitask training format uses a set of special tokens that serve as task specifiers or classification targets.
+
+
+## Setup
+
+We used Python 3.9.9 and [PyTorch](https://pytorch.org/) 1.10.1 to train and test our models, but the codebase is expected to be compatible with Python 3.7 or later and recent PyTorch versions. The codebase also depends on a few Python packages, most notably [HuggingFace Transformers](https://huggingface.co/docs/transformers/index) for their fast tokenizer implementation and [ffmpeg-python](https://github.com/kkroening/ffmpeg-python) for reading audio files. The following command will pull and install the latest commit from this repository, along with its Python dependencies 
+
+    pip install git+https://github.com/openai/whisper.git 
+
+It also requires the command-line tool [`ffmpeg`](https://ffmpeg.org/) to be installed on your system, which is available from most package managers:
+
+```bash
+# on Ubuntu or Debian
+sudo apt update && sudo apt install ffmpeg
+
+# on Arch Linux
+sudo pacman -S ffmpeg
+
+# on MacOS using Homebrew (https://brew.sh/)
+brew install ffmpeg
+
+# on Windows using Chocolatey (https://chocolatey.org/)
+choco install ffmpeg
+
+# on Windows using Scoop (https://scoop.sh/)
+scoop install ffmpeg
+```
+
+You may need [`rust`](http://rust-lang.org) installed as well, in case [tokenizers](https://pypi.org/project/tokenizers/) does not provide a pre-built wheel for your platform. If you see installation errors during the `pip install` command above, please follow the [Getting started page](https://www.rust-lang.org/learn/get-started) to install Rust development environment.
+
+
+## Available models and languages
+
+There are five model sizes, four with English-only versions, offering speed and accuracy tradeoffs. Below are the names of the available models and their approximate memory requirements and relative speed. 
+
+
+|  Size  | Parameters | English-only model | Multilingual model | Required VRAM | Relative speed |
+|:------:|:----------:|:------------------:|:------------------:|:-------------:|:--------------:|
+|  tiny  |    39 M    |     `tiny.en`      |       `tiny`       |     ~1 GB     |      ~32x      |
+|  base  |    74 M    |     `base.en`      |       `base`       |     ~1 GB     |      ~16x      |
+| small  |   244 M    |     `small.en`     |      `small`       |     ~2 GB     |      ~6x       |
+| medium |   769 M    |    `medium.en`     |      `medium`      |     ~5 GB     |      ~2x       |
+| large  |   1550 M   |        N/A         |      `large`       |    ~10 GB     |       1x       |
+
+For English-only applications, the `.en` models tend to perform better, especially for the `tiny.en` and `base.en` models. We observed that the difference becomes less significant for the `small.en` and `medium.en` models.
+
+Whisper's performance varies widely depending on the language. The figure below shows a WER breakdown by languages of Fleurs dataset, using the `large` model. More WER and BLEU scores corresponding to the other models and datasets can be found in Appendix D in [the paper](https://cdn.openai.com/papers/whisper.pdf).
+
+![WER breakdown by language](language-breakdown.svg)
+
+
+
+## Command-line usage
+
+The following command will transcribe speech in audio files, using the `medium` model:
+
+    python3 cli.py audio.wav --model medium
+
+    whisper audio.flac audio.mp3 audio.wav --model medium
+
+The default setting (which selects the `small` model) works well for transcribing English. To transcribe an audio file containing non-English speech, you can specify the language using the `--language` option:
+
+    whisper japanese.wav --language Japanese
+
+Adding `--task translate` will translate the speech into English:
+
+    whisper japanese.wav --language Japanese --task translate
+
+Run the following to view all available options:
+
+    whisper --help
+
+See [tokenizer.py](whisper/tokenizer.py) for the list of all available languages.
+
+
+## Python usage
+
+Transcription can also be performed within Python: 
+
+```python
+import whisper
+
+model = whisper.load_model("base")
+result = model.transcribe("audio.mp3")
+print(result["text"])
+```
+
+Internally, the `transcribe()` method reads the entire file and processes the audio with a sliding 30-second window, performing autoregressive sequence-to-sequence predictions on each window.
+
+Below is an example usage of `whisper.detect_language()` and `whisper.decode()` which provide lower-level access to the model.
+
+```python
+import whisper
+
+model = whisper.load_model("base")
+
+# load audio and pad/trim it to fit 30 seconds
+audio = whisper.load_audio("audio.mp3")
+audio = whisper.pad_or_trim(audio)
+
+# make log-Mel spectrogram and move to the same device as the model
+mel = whisper.log_mel_spectrogram(audio).to(model.device)
+
+# detect the spoken language
+_, probs = model.detect_language(mel)
+print(f"Detected language: {max(probs, key=probs.get)}")
+
+# decode the audio
+options = whisper.DecodingOptions()
+result = whisper.decode(model, mel, options)
+
+# print the recognized text
+print(result.text)
+```
+
+## License
+
+The code and the model weights of Whisper are released under the MIT License. See [LICENSE](LICENSE) for further details.

README.md

@@ -1,5 +1,13 @@
 # Whisper ONNX Export Script
 
+Export whisper to onnx. The decoder fixes the size of kv_cache to avoid re-allocating tensors for each inference.
+
+## Requirements
+
+- Windows or macOS or Linux
+- torch 2.0
+- onnx 1.13.1
+
 ## ONNX Export
 
 This repository based on [whisper.openvino](https://github.com/zhuzilin/whisper-openvino), but
@@ -12,149 +20,27 @@ python3 cli.py audio.wav --model medium --export_encoder
 python3 cli.py audio.wav --model medium --export_decoder
 ```
 
-You can also read weights saved_state_dicted from the original whisper.
+The following command will onnx import for inference test:
 
 ```
-python3 cli.py audio.wav --model medium --export_decoder --fine_tuning model.pth
-```
-
-The decoder fixes the size of kv_cache to avoid re-allocating tensors for each inference.
-
-## Requirements
-
-- windows or macOS or Linux
-- torch 2.0
-- onnx 1.13.1
-
-# Whisper Original information
-
-[[Blog]](https://openai.com/blog/whisper)
-[[Paper]](https://cdn.openai.com/papers/whisper.pdf)
-[[Model card]](model-card.md)
-[[Colab example]](https://colab.research.google.com/github/openai/whisper/blob/master/notebooks/LibriSpeech.ipynb)
-
-Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification.
-
-
-## Approach
-
-![Approach](approach.png)
-
-A Transformer sequence-to-sequence model is trained on various speech processing tasks, including multilingual speech recognition, speech translation, spoken language identification, and voice activity detection. All of these tasks are jointly represented as a sequence of tokens to be predicted by the decoder, allowing for a single model to replace many different stages of a traditional speech processing pipeline. The multitask training format uses a set of special tokens that serve as task specifiers or classification targets.
-
-
-## Setup
-
-We used Python 3.9.9 and [PyTorch](https://pytorch.org/) 1.10.1 to train and test our models, but the codebase is expected to be compatible with Python 3.7 or later and recent PyTorch versions. The codebase also depends on a few Python packages, most notably [HuggingFace Transformers](https://huggingface.co/docs/transformers/index) for their fast tokenizer implementation and [ffmpeg-python](https://github.com/kkroening/ffmpeg-python) for reading audio files. The following command will pull and install the latest commit from this repository, along with its Python dependencies 
-
-    pip install git+https://github.com/openai/whisper.git 
-
-It also requires the command-line tool [`ffmpeg`](https://ffmpeg.org/) to be installed on your system, which is available from most package managers:
-
-```bash
-# on Ubuntu or Debian
-sudo apt update && sudo apt install ffmpeg
-
-# on Arch Linux
-sudo pacman -S ffmpeg
-
-# on MacOS using Homebrew (https://brew.sh/)
-brew install ffmpeg
-
-# on Windows using Chocolatey (https://chocolatey.org/)
-choco install ffmpeg
-
-# on Windows using Scoop (https://scoop.sh/)
-scoop install ffmpeg
+python3 cli.py audio.wav --model medium --import_encoder
+python3 cli.py audio.wav --model medium --import_decoder
 ```
 
-You may need [`rust`](http://rust-lang.org) installed as well, in case [tokenizers](https://pypi.org/project/tokenizers/) does not provide a pre-built wheel for your platform. If you see installation errors during the `pip install` command above, please follow the [Getting started page](https://www.rust-lang.org/learn/get-started) to install Rust development environment.
-
+## ONNX Export Examples
 
-## Available models and languages
+- export.sh : Export to onnx
+- verify.sh : Verify onnx output
+- optimize.sh : Optimize onnx using ailia onnx optimizer
 
-There are five model sizes, four with English-only versions, offering speed and accuracy tradeoffs. Below are the names of the available models and their approximate memory requirements and relative speed. 
+## Export Fine Tuned Model
 
+You can also read weights saved_state_dicted from the original whisper.
 
-|  Size  | Parameters | English-only model | Multilingual model | Required VRAM | Relative speed |
-|:------:|:----------:|:------------------:|:------------------:|:-------------:|:--------------:|
-|  tiny  |    39 M    |     `tiny.en`      |       `tiny`       |     ~1 GB     |      ~32x      |
-|  base  |    74 M    |     `base.en`      |       `base`       |     ~1 GB     |      ~16x      |
-| small  |   244 M    |     `small.en`     |      `small`       |     ~2 GB     |      ~6x       |
-| medium |   769 M    |    `medium.en`     |      `medium`      |     ~5 GB     |      ~2x       |
-| large  |   1550 M   |        N/A         |      `large`       |    ~10 GB     |       1x       |
-
-For English-only applications, the `.en` models tend to perform better, especially for the `tiny.en` and `base.en` models. We observed that the difference becomes less significant for the `small.en` and `medium.en` models.
-
-Whisper's performance varies widely depending on the language. The figure below shows a WER breakdown by languages of Fleurs dataset, using the `large` model. More WER and BLEU scores corresponding to the other models and datasets can be found in Appendix D in [the paper](https://cdn.openai.com/papers/whisper.pdf).
-
-![WER breakdown by language](language-breakdown.svg)
-
-
-
-## Command-line usage
-
-The following command will transcribe speech in audio files, using the `medium` model:
-
-    python3 cli.py audio.wav --model medium
-
-    whisper audio.flac audio.mp3 audio.wav --model medium
-
-The default setting (which selects the `small` model) works well for transcribing English. To transcribe an audio file containing non-English speech, you can specify the language using the `--language` option:
-
-    whisper japanese.wav --language Japanese
-
-Adding `--task translate` will translate the speech into English:
-
-    whisper japanese.wav --language Japanese --task translate
-
-Run the following to view all available options:
-
-    whisper --help
-
-See [tokenizer.py](whisper/tokenizer.py) for the list of all available languages.
-
-
-## Python usage
-
-Transcription can also be performed within Python: 
-
-```python
-import whisper
-
-model = whisper.load_model("base")
-result = model.transcribe("audio.mp3")
-print(result["text"])
 ```
-
-Internally, the `transcribe()` method reads the entire file and processes the audio with a sliding 30-second window, performing autoregressive sequence-to-sequence predictions on each window.
-
-Below is an example usage of `whisper.detect_language()` and `whisper.decode()` which provide lower-level access to the model.
-
-```python
-import whisper
-
-model = whisper.load_model("base")
-
-# load audio and pad/trim it to fit 30 seconds
-audio = whisper.load_audio("audio.mp3")
-audio = whisper.pad_or_trim(audio)
-
-# make log-Mel spectrogram and move to the same device as the model
-mel = whisper.log_mel_spectrogram(audio).to(model.device)
-
-# detect the spoken language
-_, probs = model.detect_language(mel)
-print(f"Detected language: {max(probs, key=probs.get)}")
-
-# decode the audio
-options = whisper.DecodingOptions()
-result = whisper.decode(model, mel, options)
-
-# print the recognized text
-print(result.text)
+python3 cli.py audio.wav --model medium --export_decoder --fine_tuning model.pth
 ```
 
-## License
+# Whisper Original information
 
-The code and the model weights of Whisper are released under the MIT License. See [LICENSE](LICENSE) for further details.
+[ORIGINAL.md](ORIGINAL.md)

export.sh

@@ -0,0 +1,6 @@
+#for i in large large-v3
+for i in tiny base small medium
+do
+python3 cli.py audio.wav --model $i --export_encoder
+python3 cli.py audio.wav --model $i --export_decoder
+done

optimizer.sh

@@ -0,0 +1,11 @@
+mkdir optimize_model
+#for i in large large-v3
+for i in tiny base small medium
+do
+python3 onnx_optimizer.py export_model/encoder_${i}_opset17.onnx
+python3 onnx_optimizer.py -m optimizer/manual_opt_${i}.json export_model/decoder_${i}_opset17.onnx
+mv export_model/encoder_${i}_opset17.opt.onnx optimize_model/encoder_${i}.opt3.onnx
+mv export_model/decoder_${i}_opset17.opt.onnx optimize_model/decoder_${i}_fix_kv_cache.opt3.onnx
+python3 onnx2prototxt.py optimize_model/encoder_${i}.opt3.onnx
+python3 onnx2prototxt.py optimize_model/decoder_${i}_fix_kv_cache.opt3.onnx
+done