feat(vscode): add Parakeet speech-to-text model

This commit is contained in:
marius-kilocode
2026-06-25 14:50:00 +02:00
parent b344d17579
commit 1fb3e0679d
4 changed files with 23 additions and 2 deletions
+5
View File
@@ -0,0 +1,5 @@
---
"kilo-code": minor
---
Support NVIDIA Parakeet TDT 0.6B v3 for speech-to-text input.
@@ -38,6 +38,11 @@ const models: SpeechToTextModelDef[] = [
label: "Chirp 3",
provider: "Google",
},
{
id: "nvidia/parakeet-tdt-0.6b-v3",
label: "Parakeet TDT 0.6B v3",
provider: "NVIDIA",
},
]
export const SPEECH_TO_TEXT_MODELS: readonly SpeechToTextModelDef[] = models
@@ -207,8 +207,8 @@ test("settings and mode editing expose distinct model field purposes", async ({
const speech = page.getByRole("button", { name: "Speech to Text Model: Chirp 3" })
await expect(speech).toBeEnabled()
await speech.click()
await page.getByRole("option", { name: "GPT-4o Mini Transcribe (OpenAI)" }).click()
await expect(page.getByRole("button", { name: "Speech to Text Model: GPT-4o Mini Transcribe" })).toBeVisible()
await page.getByRole("option", { name: "Parakeet TDT 0.6B v3 (NVIDIA)" }).click()
await expect(page.getByRole("button", { name: "Speech to Text Model: Parakeet TDT 0.6B v3" })).toBeVisible()
await load(page, "settings--mode-edit-export")
await expect(page.getByRole("button", { name: /Model Override:/ })).toHaveAccessibleDescription(
@@ -14,4 +14,15 @@ describe("speech-to-text model catalog", () => {
it("falls back from unknown config model IDs", () => {
expect(getSpeechToTextModel("unknown/model")).toBe(DEFAULT_SPEECH_TO_TEXT_MODEL)
})
it("includes NVIDIA Parakeet without prompt conditioning", () => {
const model = getSpeechToTextModel("nvidia/parakeet-tdt-0.6b-v3")
expect(model).toMatchObject({
id: "nvidia/parakeet-tdt-0.6b-v3",
label: "Parakeet TDT 0.6B v3",
provider: "NVIDIA",
})
expect(model.verbatim).toBeUndefined()
})
})