*DeepSpeech2 on PaddlePaddle* is an open-source implementation of end-to-end Automatic Speech Recognition (ASR) engine, based on [Baidu's Deep Speech 2 paper](http://proceedings.mlr.press/v48/amodei16.pdf), with [PaddlePaddle](https://github.com/PaddlePaddle/Paddle) platform. Our vision is to empower both industrial application and academic research on speech recognition, via an easy-to-use, efficient and scalable implementation, including training, inferencing & testing module, distributed [PaddleCloud](https://github.com/PaddlePaddle/cloud) training, and demo deployment. Besides, several pre-trained models for both English and Mandarin are also released.
Several shell scripts provided in `./examples` will help us to quickly give it a try, for most major modules, including data preparation, model training, case inference and model evaluation, with a few public dataset (e.g. [LibriSpeech](http://www.openslr.org/12/), [Aishell](https://github.com/kaldi-asr/kaldi/tree/master/egs/aishell)). Reading these examples will also help us understand how to make it work with our own data.
Some of the scripts in `./examples` are configured with 8 GPUs. If you don't have 8 GPUs available, please modify `CUDA_VISIBLE_DEVICE` and `--trainer_count`. If you don't have any GPU available, please set `--use_gpu` to False to use CPUs instead.
Notice that this is only a toy example with a tiny sampled subset of LibriSpeech. If we would like to try with the complete dataset (would take several days for training), please go to `examples/librispeech` instead.
`run_data.sh` will download dataset, generate manifests, collect normalizer' statistics and build vocabulary. Once the data preparation is done, we will find the data (only part of LibriSpeech) downloaded in `~/.cache/paddle/dataset/speech/libri` and the corresponding manifest files generated in `./data/tiny` as well as a mean stddev file and a vocabulary file. It has to be run for the very first time we run this dataset and is reusable for all further experiments.
`run_train.sh` will start a training job, with training logs printed to stdout and model checkpoint of every pass/epoch saved to `./checkpoints/tiny`. We can resume the training from these checkpoints, or use them for inference, evaluation and deployment.
`run_infer.sh` will show us some speech-to-text decoding results for several (default: 10) samples with the trained model. The performance might not be good now as the current model is only trained with a toy subset of LibriSpeech. To see the results with a better model, we can download a well-trained (trained for several days, with the complete LibriSpeech) model and do the inference:
`run_test.sh` will evaluate the model with Word Error Rate (or Character Error Rate) measurement. Similarly, we can also download a well-trained model and test its performance:
*DeepSpeech2 on PaddlePaddle* accepts a textual **manifest** file as its data set interface. A manifest file summarizes a set of speech data, with each line containing some meta data (e.g. filepath, transcription, duration) of one audio clip, in [JSON](http://www.json.org/) format, such as:
{"audio_filepath": "/home/work/.cache/paddle/Libri/134686/1089-134686-0001.flac", "duration": 3.275, "text": "stuff it into you his belly counselled him"}
{"audio_filepath": "/home/work/.cache/paddle/Libri/134686/1089-134686-0007.flac", "duration": 4.275, "text": "a cold lucid indifference reigned in his soul"}
To use your custom data, you only need to generate such manifest files to summarize the dataset. Given such summarized manifests, training, inference and all other modules can be aware of where to access the audio files, as well as their meta data including the transcription labels.
For how to generate such manifest files, please refer to `data/librispeech/librispeech.py`, which download and generate manifests for LibriSpeech dataset.
To perform z-score normalization (zero-mean, unit stddev) upon audio features, we have to estimate in advance the mean and standard deviation of the features, with some training samples:
It will compute the mean and standard deviation of power spectrum feature with 2000 random sampled audio clips listed in `data/librispeech/manifest.train` and save the results to `data/librispeech/mean_std.npz` for further usage.
A vocabulary of possible characters is required to convert the transcription into a list of token indices for training, and in decoding, to convert from a list of indices back to text again. Such a character-based vocabulary can be built with `tools/build_vocab.py`.
It will write a vocabuary file `data/librispeeech/eng_vocab.txt` with all transcription text in `data/librispeech/manifest.train`, without vocabulary truncation (`--count_threshold 0`).
Data augmentation has often been a highly effective technique to boost the deep learning performance. We augment our speech data by synthesizing new audios with small random perturbation (label-invariant transformation) added upon raw audios. We don't have to do the syntheses by ourselves, as it is already embedded into the data provider and is done on the fly, randomly for each epoch during training.
In order to inform the trainer of what augmentation components we need and what their processing orders are, we are required to prepare a *augmentation configuration file* in [JSON](http://www.json.org/) format. For example:
When the `--augment_conf_file` argument of `trainer.py` is set to the path of the above example configuration file, every audio clip in every epoch will be processed: with 60% of chance, it will first be speed perturbed with a uniformly random sampled speed-rate between 0.95 and 1.05, and then with 80% of chance it will be shifted in time with a random sampled offset between -5 ms and 5 ms. Finally this newly synthesized audio clip will be feed into the feature extractor for further training.
Be careful when we are utilizing the data augmentation technique, as improper augmentation will do harm to the training, due to the enlarged train-test gap.
A language model is required to improve the decoder's performance. We have prepared two language models (with lossy compression) for users to download and try. One is for English and the other is for Mandarin. We can simply run this to download the preprared language models:
An inference module caller `infer.py` is provided for us to infer, decode and visualize speech-to-text results for several given audio clips. It might help to have an intuitive and qualitative evaluation of the ASR model's performance.
We provide two types of CTC decoders: *CTC greedy decoder* and *CTC beam search decoder*. The *CTC greedy decoder* is an implementation of the simple best-path decoding algorithm, selecting at each timestep the most likely token, thus being greedy and locally optimal. The [*CTC beam search decoder*](https://arxiv.org/abs/1408.2873) otherwise utilizes a heuristic breadth-first graph search for reaching a near global optimality; it also requires a pre-trained KenLM language model for better scoring and ranking. The decoder type can be set with argument `--decoding_method`.
The hyper-parameters $\alpha$ (coefficient for language model scorer) and $\beta$ (coefficient for word count scorer) for the [*CTC beam search decoder*](https://arxiv.org/abs/1408.2873) often have a significant impact on the decoder's performance. It would be better to re-tune them on a validation set when the acoustic model is renewed.
`tools/tune.py` performs a 2-D grid search over the hyper-parameter $\alpha$ and $\beta$. We must provide the range of $\alpha$ and $\beta$, as well as the number of their attempts.
We provide a cloud training module for users to do the distributed cluster training on [PaddleCloud](https://github.com/PaddlePaddle/cloud), to achieve a much faster training speed with multiple machines. To start with this, please first install PaddleCloud client and register a PaddleCloud account, as described in [PaddleCloud Usage](https://github.com/PaddlePaddle/cloud/blob/develop/doc/usage_cn.md#%E4%B8%8B%E8%BD%BD%E5%B9%B6%E9%85%8D%E7%BD%AEpaddlecloud).
Given input manifests, `pcloud_upload_data.sh` will:
- Extract the audio files listed in the input manifests.
- Pack them into a specified number of tar files.
- Upload these tar files to PaddleCloud filesystem.
- Create cloud manifests by replacing local filesystem paths with PaddleCloud filesystem paths. New manifests will be used to inform the cloud jobs of audio files' location and their meta information.
It should be done only once for the very first time we do the cloud training. Later, the data is kept persisitent on the cloud filesystem and reusable for further job submissions.
For argument details please refer to [Train DeepSpeech2 on PaddleCloud](https://github.com/PaddlePaddle/models/tree/develop/deep_speech_2/cloud).
- Configure training arguments:
Configure the cloud job parameters in `pcloud_submit.sh` (e.g. `NUM_NODES`, `NUM_GPUS`, `CLOUD_TRAIN_DIR`, `JOB_NAME` etc.) and then configure other hyper-parameters for training in `pcloud_train.sh` (just as what you do for local training).
For argument details please refer to [Train DeepSpeech2 on PaddleCloud](https://github.com/PaddlePaddle/models/tree/develop/deep_speech_2/cloud).
we submit a training job to PaddleCloud. And the job name will be printed when the submission is finished. Now our training job is running well on the PaddleCloud.
For more information about the usage of PaddleCloud, please refer to [PaddleCloud Usage](https://github.com/PaddlePaddle/cloud/blob/develop/doc/usage_cn.md#提交任务).
For more information about the DeepSpeech2 training on PaddleCloud, please refer to
Until now, we have trained and tested our ASR model qualitatively (`infer.py`) and quantitatively (`test.py`) with existing audio files. But we have not yet try the model with our own speech. `deploy/demo_server.py` and `deploy/demo_client.py` helps quickly build up a real-time demo ASR engine with the trained model, enabling us to test and play around with the demo, with our own voice.
Now, in the client console, press the `whitespace` key, hold, and start speaking. Until we finish our utterance, we release the key to let the speech-to-text results shown in the console. To quit the client, just press `ESC` key.
Notice that `deploy/demo_client.py` must be run in a machine with a microphone device, while `deploy/demo_server.py` could be run in one without any audio recording hardware, e.g. any remote server machine. Just be careful to set the `host_ip` and `host_port` argument with the actual accessible IP address and port, if the server and client are running with two separate machines. Nothing should be done if they are running in one single machine.
We can also refer to `examples/mandarin/run_demo_server.sh` for example, which will first download a pre-trained Mandarin model (trained with 3000 hours of internal speech data) and then start the demo server with the model. With running `examples/mandarin/run_demo_client.sh`, we can speak Mandarin to test it. If we would like to try some other models, just update `--model_path` argument in the script.
You are welcome to submit questions and bug reports in [Github Issues](https://github.com/PaddlePaddle/models/issues). You are also welcome to contribute to this project.