You need to include the siteId as well as the text.
Are you watching the snips mqtt with snips_watch -vvv? Will show the messages as they come across
you can test using mosquitto_pub, part of the mosquitto-clients package usually.
mosquitto_pub -h SNIPS_MQTT_IP -p 1883 -t hermes/tts/say -m “{"siteId":"default","text":"hi there"}”
Note, you can also play a wav file directly to the audio server. With this you could get the file cached from the google say or something.
mosquitto_pub -h SNIPS_MQTT_IP -t ‘hermes/audioServer/default/playBytes/0049a91e-8449-4398-9752-07c0e1858234’ -f /usr/share/sounds/alsa/Front_Center.wav
You can see how I set it up here.