NTT Corevo Speech Recognition

This is an unofficial git repo for a Docker image of NTT's Corevo Speech Recognition SDK build environment.

To build the image:

$ cd ntt/
$ docker build -t ntt-corevo-asr .

The Dockerfile requires the AsrTestDriver.sh file, which also has the user's API credentials. This file is not included here for security reasons. Go to NTT Corevo API Portal to get your credentials and download the Speech Recognition SDK. The AsrTestDriver.sh file is located in corevo_ASR_SDK_for_Linux_v1.0.1.0/testDriver/bin/.

Run the default example:

$ docker run ntt-corevo-asr

This runs the AsrTestDriver.sh file, which by default transcribes the ./pcm/ja_short.pcm file that is already in the container.

Run a custom voice sample:

$ docker run -v /host/path:/container/path ntt-corevo-asr -f /container/path/<filename>.wav

where

• /host/path is the path to the folder containing the .wav file on the host machine
• /container/path is a unique path in the container that will be used to mount the /host/path

For example:

$ docker run -v /home/rayhrst/data/:/home/data/ ntt-corevo-asr -f /home/data/input.wav