Hey all, new to the forum here. I am looking at implementing a custom TTS engine. Yes, I know there is a long list of supported engines, but I am attempting to use eleven labs specifically. I am at a point where (I think) I implemented all the architecture necessary, but for some reason the TTS simply is not working. I verified that the problem essentially rests either in the tts file itself, or on the mycroft side of things. Here is my mycroft conf file:
{ "max_allowed_core_version": 21.2, "tts": { "elevenlabs": { "voice_id": "voice id here", "api_key": "my api key here" }, "module": "elevenlabs" } }
and my elevenlabs_tts.py:
`
import requests
from .tts import TTS, TTSValidator
from mycroft.configuration import Configuration
_API_URL = âhttps://api.elevenlabs.io/v1/text-to-speechâ
class ElevenLabsTTS(TTS):
def init(self, lang, config):
super(ElevenLabsTTS, self).init(lang, config, ElevenLabsTTSValidator(self))
self.config = Configuration.get().get(âttsâ, {}).get(âelevenlabsâ, {})
self.voice_id = self.config.get(âvoice_idâ)
self.api_key = self.config.get(âapi_keyâ)
self.type = âmp3â
def get_tts(self, sentence, mp3_file):
url = "{}/{}".format(_API_URL, self.voice_id)
headers = {
"Accept": "audio/mpeg",
"Content-Type": "application/json",
"xi-api-key": self.api_key
}
data = {
"text": sentence,
"voice_settings": {
"stability": 0,
"similarity_boost": 0
}
}
response = requests.post(url, json=data, headers=headers)
CHUNK_SIZE = 1024
with open(mp3_file, 'wb') as f:
for chunk in response.iter_content(chunk_size=CHUNK_SIZE):
if chunk:
f.write(chunk)
return (mp3_file, None) # No phonemes
class ElevenLabsTTSValidator(TTSValidator):
def init(self, tts):
super(ElevenLabsTTSValidator, self).init(tts)
def validate_lang(self):
# Assuming Eleven Labs API supports the language set in Mycroft
return True
def validate_connection(self):
# Assuming the API key and voice ID are correct
return True
def get_tts_class(self):
return ElevenLabsTTS`
I have verified that the api works in a separate Python example, so there isnât something wrong with the api key or voice id. Also, I included the file my importing in tts.py:
from mycroft.tts.elevenlabs_tts import ElevenLabsTTS
and
"elevenlabs": ElevenLabsTTS â under CLASSES
For some reason, still, I cannot get mycroft to use the custom module. It seems like this should work considering the same implementation for other files, but something is clearly missing.