The Speechify TTS plugin page in the docs (https://docs.livekit.io/agents/models/tts/plugins/speechify) shows this example:
tts=speechify.TTS(
model="simba-english",
voice_id="jack",
)
simba-english (and simba-multilingual, also listed on that page) are Speechify's Simba 1.6 models. They are retired for new integrations from Speechify API version 2026-09-21, so a developer on a new account who copies this example gets a 400 model_retired on the first call. From 2026-11-21 both ids are served by Speechify's current models (Speechify's models guide: https://docs.speechify.ai/build/guides/concepts/models#simba-16-and-your-api-version).
The plugin itself is already current: livekit-plugins/livekit-plugins-speechify/livekit/plugins/speechify/tts.py defaults to DEFAULT_MODEL = "simba-3.2" and DEFAULT_VOICE_ID = "dominic_32". Only the docs page lags.
Suggested fix for the docs example:
tts=speechify.TTS(
model="simba-3.2",
voice_id="dominic_32",
)
with simba-3.0 named as the model for non-English voices, and the two Simba 1.6 ids removed from the model list or marked retired. The plugin also produces word timestamps on the streaming models (simba-3.0 and simba-3.2), which the page could mention.
Happy to help with the wording. I work on the Speechify API.
The Speechify TTS plugin page in the docs (https://docs.livekit.io/agents/models/tts/plugins/speechify) shows this example:
simba-english(andsimba-multilingual, also listed on that page) are Speechify's Simba 1.6 models. They are retired for new integrations from Speechify API version2026-09-21, so a developer on a new account who copies this example gets a400 model_retiredon the first call. From 2026-11-21 both ids are served by Speechify's current models (Speechify's models guide: https://docs.speechify.ai/build/guides/concepts/models#simba-16-and-your-api-version).The plugin itself is already current:
livekit-plugins/livekit-plugins-speechify/livekit/plugins/speechify/tts.pydefaults toDEFAULT_MODEL = "simba-3.2"andDEFAULT_VOICE_ID = "dominic_32". Only the docs page lags.Suggested fix for the docs example:
with
simba-3.0named as the model for non-English voices, and the two Simba 1.6 ids removed from the model list or marked retired. The plugin also produces word timestamps on the streaming models (simba-3.0andsimba-3.2), which the page could mention.Happy to help with the wording. I work on the Speechify API.