Speech to text
5.23'22
vosk
Vosk Server
Server: requires about 8G
memory or more.
docker run -d -p 2700:2700 alphacep/kaldi-en:latest
Client
git clone https://github.com/alphacep/vosk-server
cd vosk-server/websocket
./test.py test.wav
Client with microphone
pip3 install sounddevice
./test_microphone.py -u ws://localhost:2700
Julius
https://github.com/julius-speech/julius
Build
git clone https://github.com/julius-speech/julius.git
cd julius
Fix for macOS: Add belowto libsent/src/adin_mic_darwin_coreaudio.c
#include <sent/stddefs.h>
./configure --enable-words-int
make -j4
Run
Download JuliusModel named ENVR-v5.4.Dnn.Bin.zip
unzip ENVR-v5.4.Dnn.Bin.zip
cd ENVR-v5.4.Dnn.Bin
Edit config file dnn.jconf
-feature_options -htkconf wav_config -cvn -cmnload ENVR-v5.3.norm -cmnstatic
+feature_options -htkconf wav_config -cvn -cmnload ENVR-v5.3.norm -cvnstatic
+state_prior_log10nize false
Recognize audio file
../julius/julius/julius -C julius.jconf -dnnconf dnn.jconf
Edit test.dbl
to test with other audio files.
📖