Skip to content

Latest commit

 

History

History
36 lines (21 loc) · 1.93 KB

README.md

File metadata and controls

36 lines (21 loc) · 1.93 KB
Raw

NOTE:

This directory contains two sub-directories one for WINDOWS OS and one for LINUX OS. Read corresponding READMEs for each OS .

Only difference for both OS is the library used for gathering audio data from microphone .On WINDOWS portaudio is used while on LINUX ALSA-lib C is used which itself provides an interface to ALSA Kernel module.

Interface to both the libs is provided through NIM code.

PREREQUISITIES :

  • libdeepspeech.so

Go to the releases page and download the native client package based on your OS and CPU architecture.

Extract the libdeepspeech.so and put into the subdirectory depending on OS of native Client used.

On WINDOWS:

  • Download the native.client.amd64.win.tar.xz package . [ same is true for xx.xx.amd64.cuda.win.xx if CUDA installed or xx.xx.amd64.tflite.win.xx]
  • Extract and place the libdeepspeech.so in win_nim_vad_streaming subdirectory
  • Now see README.md in win_nim_vad_streaming subdirectory.

On LINUX:

  • Download the native_client.amd64.linux.cpu package .[ same is true for xx.xx.amd64.cuda.linux.xx is CUDA installed or xx.xx.amd64.tflite.linux.xx]
  • Extract and place the libdeepspeech.so in linux_nim_vad_streaming subdirectory
  • Now see README.md in linux_nim_vad_streaming subdirectory.

Note: One can put libdeepspeech.so in the system's PATH rather than copying it to one of subdirectories for easy usage.

NOTE:

Used NIM code only depends on the shared library(libdeepspeech.so) used. Given one has downloaded the native client package and extracted the libdeepspeech.so shared library and copied it to one of the subdirectories or in system's PATH ,Code can be modified to add more functionalities in pure NIM and modified code would compile on any platform as long as that platform is supported by NIM.