We are using it on Android devices. Initially, we used the webSocket server. If the speaker is far away from the device, about 1 meter away, it will be difficult or unrecognizable.
Later, we switched to using the webRtc server, and the effect improved slightly, but it was still not good.
Has anyone encountered a similar situation before?
Is there any way to improve the accuracy of long-distance recognition?
You have to find long-distance model and also you need a very good microphone. Not sure what device are you using but most likely you need a special device.
We are using it on Android devices. Initially, we used the webSocket server. If the speaker is far away from the device, about 1 meter away, it will be difficult or unrecognizable. Later, we switched to using the webRtc server, and the effect improved slightly, but it was still not good. Has anyone encountered a similar situation before? Is there any way to improve the accuracy of long-distance recognition?