meta-offline-voice-agent/README

   1 Offline Speech Recognition using Vosk (Kaldi-ASR) : meta-offline-voice-agent
   2 =========================================================================
   3
   4 meta-offline-voice-agent is the AGL Layer using Vosk API based on Kaldi ASR Toolkit to enable Offline Speech Recognition capabilities for Automotive Grade Linux.
   5
   6
   7 WIP
   8 ========
   9 The layer in its current state contains the Vosk library, and is capable of recognizing speech, as verified from the test scripts contained in https://github.com/alphacep/vosk-api/tree/master/python/example.
  10
  11 ### Working features:
  12 - [vosk-api (python)](https://github.com/alphacep/vosk-api/tree/master/python)
  13 - [vosk-websocket-server](https://github.com/alphacep/vosk-server/tree/master/websocket)
  14
  15
  16 Testing vosk-api on AGL
  17 ======================
  18
  19 ### 1. Initializing the build environment:
  20
  21 The `agl-offline-voice-agent` feature needs to be enabled when including aglsetup.sh:
  22
  23 ```shell
  24 $ source meta-agl/scripts/aglsetup.sh -m qemux86-64 -b build-voice-qemux86-64 agl-demo agl-offline-voice-agent ${AGL_META_PYTHON}
  25
  26 $ bitbake agl-demo-platform
  27 ```
  28
  29 ### 2. Running the image:
  30
  31 ```shell
  32 $ runqemu tmp/deploy/images/qemux86-64/agl-demo-platform-qemux86-64.qemuboot.conf kvm serialstdio slirp publicvnc audio
  33 ```
  34
  35 ### 3. Run the test with ptest-runner:
  36
  37 ```shell
  38 $ ptest-runner python3-vosk-api
  39 ```
  40
  41 ### Currently supported targets:
  42 - QEMU x86-64: Work in progress.
  43
  44 Maintainers:
  45 - Aman Arora <aman.arora9848@gmail.com>
  46