You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
PaddleSpeech/speechx/README.md

63 lines
1.8 KiB

3 years ago
# SpeechX -- All in One Speech Task Inference
3 years ago
3 years ago
## Environment
We develop under:
* docker - `registry.baidubce.com/paddlepaddle/paddle:2.2.2-gpu-cuda10.2-cudnn7`
3 years ago
* os - Ubuntu 16.04.7 LTS
* gcc/g++/gfortran - 8.2.0
3 years ago
* cmake - 3.16.0
> We make sure all things work fun under docker, and recommend using it to develop and deploy.
* [How to Install Docker](https://docs.docker.com/engine/install/)
* [A Docker Tutorial for Beginners](https://docker-curriculum.com/)
* [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/overview.html)
3 years ago
## Build
3 years ago
1. First to launch docker container.
```
docker run --privileged --net=host --ipc=host -it --rm -v $PWD:/workspace --name=dev registry.baidubce.com/paddlepaddle/paddle:2.2.2-gpu-cuda10.2-cudnn7 /bin/bash
3 years ago
```
3 years ago
* More `Paddle` docker images you can see [here](https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/install/docker/linux-docker.html).
2. Build `speechx` and `examples`.
> Do not source venv.
3 years ago
```
pushd /path/to/speechx
3 years ago
./build.sh
3 years ago
```
3 years ago
3 years ago
3. Go to `examples` to have a fun.
More details please see `README.md` under `examples`.
## Valgrind (Optional)
3 years ago
> If using docker please check `--privileged` is set when `docker run`.
3 years ago
* Fatal error at startup: `a function redirection which is mandatory for this platform-tool combination cannot be set up`
3 years ago
```
apt-get install libc6-dbg
```
3 years ago
* Install
3 years ago
```
pushd tools
./setup_valgrind.sh
popd
```
3 years ago
## TODO
### Deepspeech2 with linear feature
3 years ago
* DecibelNormalizer: there is a little bit difference between offline and online db norm. The computation of online db norm read feature chunk by chunk, which causes the feature size is different with offline db norm. In normalizer.cc:73, the samples.size() is different, which causes the difference of result.