typing `get_tts_model_voices`

[cqjjjzr]

Currently, TTSModel.get_tts_model_voices returns Any and I didn't find any clues indicating the concrete code producing the return value of get_tts_model_voices.

graphon/src/graphon/model_runtime/protocols/tts_runtime.py

Lines 26 to 33 in bfe73c1

	def get_tts_model_voices(
	self,
	*,
	provider: str,
	model: str,
	credentials: dict[str, Any],
	language: str | None,
	) -> Any: ...

How should we type this function's return value?

[cqjjjzr]

Found the concrete code at Dify's code:

https://github.com/langgenius/dify/blob/a8236499343f34cdb665b80c90296f221bc16d60/api/core/plugin/impl/model.py#L503-L545

[dosubot]

Based on the Dify implementation you linked, get_tts_model_voices returns list[dict[str, str]] — each dict has "name" and "value" keys [1]:

voices.append({"name": voice.name, "value": voice.value})

There's currently no TypedDict or type alias in graphon for this. One option would be to introduce a TypedDict like:

class TTSModelVoice(TypedDict):
    name: str
    value: str

and type the return as list[TTSModelVoice].

Worth noting: the graphon test stubs currently return list[str] (just voice names) rather than the structured dict format [2], so those would need updating to match.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.