feat(server): list experimental voice models (#57194)

This commit is contained in:
Kit Langton
2026-10-20 23:13:14 -07:00
committed by GitHub
Unverified
parent a2cb1c4e7f
commit 79fd806343
16 changed files with 634 additions and 75 deletions
+1
View File
@@ -33,4 +33,5 @@ export type {
RealtimeSession,
RealtimeTransport,
RealtimeTransportInput,
RealtimeTransportType,
} from "./experimental/realtime.js"
+23 -13
View File
@@ -12,6 +12,8 @@ export type RealtimeTransportInput =
| { readonly type: "webrtc"; readonly sdp: string; readonly dcid?: number }
| { readonly type: "websocket" }
export type RealtimeTransportType = RealtimeTransportInput["type"]
export type RealtimeTransport =
| { readonly type: "webrtc"; readonly sdp: string }
| { readonly type: "websocket"; readonly url: string; readonly token: string; readonly expiresAt: number }
@@ -25,11 +27,23 @@ export interface RealtimeConnectInput {
readonly transport: RealtimeTransportInput
}
const engines = [
{ family: "gpt-live", id: "openai-live", provider: "openai" },
{ family: "gpt-realtime", id: "openai-realtime", provider: "openai" },
{ family: "grok-voice", id: "xai-realtime", provider: "xai" },
{ family: "gemini-3.8-live", id: "gemini-live", provider: "google" },
export const engines = [
{ family: "gpt-live", model: "gpt-live-1", id: "openai-live", provider: "openai", transports: ["webrtc"] },
{
family: "gpt-realtime",
model: "gpt-realtime-2.1",
id: "openai-realtime",
provider: "openai",
transports: ["webrtc", "websocket"],
},
{ family: "grok-voice", model: "grok-voice-3.0", id: "xai-realtime", provider: "xai", transports: ["websocket"] },
{
family: "gemini-3.8-live",
model: "gemini-3.8-live",
id: "gemini-live",
provider: "google",
transports: ["websocket"],
},
] as const
// Matches the bare family name (`gpt-realtime`) and its versions (`gpt-realtime-2.1`, `gpt-realtime-mini`).
@@ -39,12 +53,12 @@ export const engine = (model: string) =>
export const connect = Effect.fn("Realtime.connect")(function* (input: RealtimeConnectInput) {
const resolved = engine(input.model)
if (!resolved) return yield* invalidRequest(`Unsupported voice model: ${input.model}`)
if (!resolved.transports.some((type) => type === input.transport.type))
return yield* invalidRequest(`${input.model} only supports ${resolved.transports.join(", ")} transport`)
const session = { ...input.session, model: input.model }
switch (resolved.id) {
case "openai-live": {
if (input.transport.type !== "webrtc")
return yield* invalidRequest("OpenAI GPT-Live only supports WebRTC transport")
const data = yield* post(
resolved.id,
`${OPENAI_BASE_URL}/live/sessions`,
@@ -73,16 +87,11 @@ export const connect = Effect.fn("Realtime.connect")(function* (input: RealtimeC
)
return { type: "webrtc" as const, sdp }
}
case "xai-realtime": {
if (input.transport.type !== "websocket")
return yield* invalidRequest("xAI Grok Voice only supports WebSocket transport")
case "xai-realtime":
return yield* clientSecret(resolved.id, XAI_BASE_URL, input.model, input.apiKey, {
expires_after: { seconds: 60 },
})
}
case "gemini-live": {
if (input.transport.type !== "websocket")
return yield* invalidRequest("Google Gemini Live only supports WebSocket transport")
const now = yield* Clock.currentTimeMillis
const expiresAt = now + 60_000
const data = yield* post(
@@ -109,6 +118,7 @@ export const connect = Effect.fn("Realtime.connect")(function* (input: RealtimeC
})
export const Realtime = {
engines,
engine,
connect,
} as const
+3
View File
@@ -138,6 +138,9 @@ describe("experimental Realtime", () => {
)
test("routes bare and versioned model names to their engine", () => {
expect(Realtime.engines.map((item) => Realtime.engine(item.model)?.id)).toEqual(
Realtime.engines.map((item) => item.id),
)
expect(
[
"gpt-realtime",
+12
View File
@@ -1418,6 +1418,17 @@ export interface GenerateApi<E = never> {
readonly text: GenerateTextOperation<E>
}
export type VoiceModelsOutput = {
readonly models: ReadonlyArray<{
readonly model: string
readonly provider: string
readonly transports: ReadonlyArray<"webrtc" | "websocket">
readonly available: boolean
}>
readonly default?: string | undefined
}
export type VoiceModelsOperation<E = never> = () => Effect.Effect<VoiceModelsOutput, E>
export type VoiceSessionInput = {
readonly model: string
readonly session?: { readonly [x: string]: unknown } | undefined
@@ -1433,6 +1444,7 @@ export type VoiceSessionOutput = {
export type VoiceSessionOperation<E = never> = (input: VoiceSessionInput) => Effect.Effect<VoiceSessionOutput, E>
export interface VoiceApi<E = never> {
readonly models: VoiceModelsOperation<E>
readonly session: VoiceSessionOperation<E>
}
+13 -1
View File
@@ -115,6 +115,7 @@ import type {
ModelDefaultOutput,
GenerateTextInput,
GenerateTextOutput,
VoiceModelsOutput,
VoiceSessionInput,
VoiceSessionOutput,
ProviderListInput,
@@ -873,6 +874,14 @@ const EndpointGenerateText = (raw: RawClient["server.generate"]) => (input: Gene
const adaptGroupGenerate = (raw: RawClient["server.generate"]) => ({ text: EndpointGenerateText(raw) })
const EndpointVoiceModels = (raw: RawClient["server.voice"]) => () =>
preserveEffect<VoiceModelsOutput>()(
raw["voice.models"]({}).pipe(
Effect.mapError(mapClientError),
Effect.map((value) => value.data),
),
)
const EndpointVoiceSession = (raw: RawClient["server.voice"]) => (input: VoiceSessionInput) =>
preserveEffect<VoiceSessionOutput>()(
raw["voice.session"]({
@@ -883,7 +892,10 @@ const EndpointVoiceSession = (raw: RawClient["server.voice"]) => (input: VoiceSe
),
)
const adaptGroupVoice = (raw: RawClient["server.voice"]) => ({ session: EndpointVoiceSession(raw) })
const adaptGroupVoice = (raw: RawClient["server.voice"]) => ({
models: EndpointVoiceModels(raw),
session: EndpointVoiceSession(raw),
})
const EndpointProviderList = (raw: RawClient["server.provider"]) => (input?: ProviderListInput) =>
preserveEffect<ProviderListOutput>()(
@@ -109,6 +109,7 @@ import type {
ModelDefaultOutput,
GenerateTextInput,
GenerateTextOutput,
VoiceModelsOutput,
VoiceSessionInput,
VoiceSessionOutput,
ProviderListInput,
@@ -1162,6 +1163,17 @@ export function make(options: ClientOptions) {
).then((value) => value.data),
},
voice: {
models: (requestOptions?: RequestOptions) =>
request<{ readonly data: VoiceModelsOutput }>(
{
method: "GET",
path: `/api/experimental/voice/models`,
successStatus: 200,
declaredStatuses: [400, 401, 404, 503],
empty: false,
},
requestOptions,
).then((value) => value.data),
session: (input: VoiceSessionInput, requestOptions?: RequestOptions) =>
request<{ readonly data: VoiceSessionOutput }>(
{
@@ -180,6 +180,8 @@ export type MoneyUSDPerMillionTokens = number
export type GenerateTextResponse = { data: { text: string } }
export type VoiceTransportType = "webrtc" | "websocket"
export type VoiceWebRTCTransport = { type: "webrtc"; sdp: string }
export type VoiceWebSocketTransport = { type: "websocket"; url: string; token: string; expiresAt: number }
@@ -662,6 +664,8 @@ export type ModelCost = {
cache: { read: MoneyUSDPerMillionTokens; write: MoneyUSDPerMillionTokens }
}
export type VoiceModel = { model: string; provider: string; transports: Array<VoiceTransportType>; available: boolean }
export type VoiceTransport = VoiceWebRTCTransport | VoiceWebSocketTransport
export type IntegrationCommandMethod = {
@@ -1985,6 +1989,8 @@ export type ProviderInfo = {
body?: { [x: string]: any }
}
export type VoiceModelsResponse = { data: { models: Array<VoiceModel>; default?: string | undefined } }
export type VoiceSessionResponse = { data: { transport: VoiceTransport } }
export type CredentialValue = CredentialOAuth | CredentialKey | CredentialExternal
@@ -4966,6 +4972,8 @@ export type GenerateTextInput = {
export type GenerateTextOutput = GenerateTextResponse["data"]
export type VoiceModelsOutput = VoiceModelsResponse["data"]
export type VoiceSessionInput = {
readonly model: {
readonly model: string
+1 -1
View File
@@ -46,7 +46,7 @@ test("exposes every standard HTTP API group", () => {
expect(Object.keys(client.integration.oauth)).toEqual(["connect", "status", "complete", "cancel"])
expect(Object.keys(client.integration.command)).toEqual(["connect", "status", "cancel"])
expect(Object.keys(client.websearch)).toEqual(["providers", "query"])
expect(Object.keys(client.voice)).toEqual(["session"])
expect(Object.keys(client.voice)).toEqual(["models", "session"])
expect(Object.keys(client.file)).toEqual(["read", "list", "find", "write"])
expect(Object.keys(client.vcs)).toEqual(["init", "get", "base", "status", "branch", "diff"])
expect(Object.keys(client.pty)).toEqual(["list", "create", "get", "update", "remove", "connect"])
+37 -5
View File
@@ -6,6 +6,7 @@ import {
type RealtimeSession,
type RealtimeTransport,
type RealtimeTransportInput,
type RealtimeTransportType,
} from "@opencode/ai/experimental"
import { RequestExecutor } from "@opencode/ai/route"
import { makeLocationNode } from "@opencode/util/effect/app-node"
@@ -13,6 +14,18 @@ import { Context, Effect, Layer, Redacted, Schema } from "effect"
import { requestExecutor } from "./effect/app-node-platform.js"
import { Integration } from "./integration.js"
export interface ModelInfo {
readonly model: string
readonly provider: string
readonly transports: readonly RealtimeTransportType[]
readonly available: boolean
}
export interface Models {
readonly models: readonly ModelInfo[]
readonly default?: string
}
export interface SessionInput {
readonly model: string
readonly session?: RealtimeSession
@@ -31,6 +44,7 @@ export class UnavailableError extends Schema.TaggedError<UnavailableError>()("Vo
export type Error = InvalidRequestError | UnavailableError
export interface Interface {
readonly models: () => Effect.Effect<Models>
readonly session: (input: SessionInput) => Effect.Effect<RealtimeTransport, Error>
}
@@ -42,14 +56,32 @@ export const layer = Layer.effect(
const executor = yield* RequestExecutor.Service
const integrations = yield* Integration.Service
// Integration lists the active credential first, then newer stored credentials, then non-empty env vars.
const findConnection = (provider: string) =>
integrations
.get(Integration.ID.make(provider))
.pipe(Effect.map((info) => info?.connections.find((item) => item.type === "env" || item.method === "key")))
const models = Effect.fn("Voice.models")(function* () {
const list = yield* Effect.forEach(Realtime.engines, (engine) =>
findConnection(engine.provider).pipe(
Effect.map((connection) => ({
model: engine.model,
provider: engine.provider,
transports: engine.transports,
available: Boolean(connection),
})),
),
)
const first = list.find((item) => item.available)?.model
return first ? { models: list, default: first } : { models: list }
})
const session = Effect.fn("Voice.session")(function* (input: SessionInput) {
const engine = Realtime.engine(input.model)
if (!engine) return yield* new InvalidRequestError({ message: `Unsupported voice model: ${input.model}` })
// Integration lists the active credential first, then newer stored credentials, then its env vars.
const connection = (yield* integrations.get(Integration.ID.make(engine.provider)))?.connections.find(
(item) => item.type === "env" || item.method === "key",
)
const connection = yield* findConnection(engine.provider)
const credential = connection
? yield* integrations.connection
.resolve(connection)
@@ -74,7 +106,7 @@ export const layer = Layer.effect(
)
})
return Service.of({ session })
return Service.of({ models, session })
}),
)
+80 -33
View File
@@ -24,6 +24,29 @@ const keys: Record<string, string> = {
GEMINI_API_KEY: "gemini-env-key",
}
const unused = () => Effect.die("unused")
const mockIntegration = (
connections: (id: Integration.ID) => Integration.Info["connections"],
resolve: Integration.Interface["connection"]["resolve"] = () => Effect.die("Voice.models resolved a credential"),
) =>
Layer.mock(Integration.Service, {
revision: () => 0,
get: (id) => Effect.succeed({ id, name: id, methods: [], connections: connections(id) }),
connection: {
active: unused,
resolve,
key: unused,
external: unused,
activate: unused,
update: unused,
remove: unused,
status: unused,
},
oauth: { connect: unused, status: unused, complete: unused, cancel: unused },
command: { connect: unused, status: unused, cancel: unused },
})
it.effect("uses the first key or env connection Integration lists and never a ChatGPT sign-in", () =>
Effect.gen(function* () {
yield* TestClock.setTime(1760000000_000)
@@ -42,39 +65,18 @@ it.effect("uses the first key or env connection Integration lists and never a Ch
Voice.layer.pipe(
Layer.provide(
Layer.merge(
Layer.mock(Integration.Service, {
revision: () => 0,
get: (id) => Effect.succeed({ id, name: id, methods: [], connections }),
connection: {
active: () => Effect.die("unused"),
resolve: (connection) =>
connection.type === "credential" && connection.method === "oauth"
? Effect.die("a ChatGPT sign-in was resolved")
: Effect.succeed(
Credential.Key.make({
type: "key",
key: keys[connection.type === "env" ? connection.name : connection.id],
}),
),
key: () => Effect.die("unused"),
external: () => Effect.die("unused"),
activate: () => Effect.die("unused"),
update: () => Effect.die("unused"),
remove: () => Effect.die("unused"),
status: () => Effect.die("unused"),
},
oauth: {
connect: () => Effect.die("unused"),
status: () => Effect.die("unused"),
complete: () => Effect.die("unused"),
cancel: () => Effect.die("unused"),
},
command: {
connect: () => Effect.die("unused"),
status: () => Effect.die("unused"),
cancel: () => Effect.die("unused"),
},
}),
mockIntegration(
() => connections,
(connection) =>
connection.type === "credential" && connection.method === "oauth"
? Effect.die("a ChatGPT sign-in was resolved")
: Effect.succeed(
Credential.Key.make({
type: "key",
key: keys[connection.type === "env" ? connection.name : connection.id],
}),
),
),
RequestExecutor.layer.pipe(
Layer.provide(
Layer.succeed(
@@ -128,3 +130,48 @@ it.effect("uses the first key or env connection Integration lists and never a Ch
expect(missingGoogle.message).toContain("GEMINI_API_KEY")
}),
)
it.effect("lists voice models with availability and default without resolving credentials", () =>
Effect.gen(function* () {
const list = (connections: Partial<Record<string, Integration.Info["connections"]>>) =>
Effect.flatMap(Voice.Service, (voice) => voice.models()).pipe(
Effect.provide(
Voice.layer.pipe(
Layer.provide(
Layer.merge(
mockIntegration((id) => connections[id] ?? []),
Layer.mock(RequestExecutor.Service, {}),
),
),
),
),
)
expect(
yield* list({
openai: [credential("chatgpt", "oauth")],
xai: [{ type: "env", name: "XAI_API_KEY" }],
google: [credential("gemini", "key")],
}),
).toEqual({
models: [
{ model: "gpt-live-1", provider: "openai", transports: ["webrtc"], available: false },
{ model: "gpt-realtime-2.1", provider: "openai", transports: ["webrtc", "websocket"], available: false },
{ model: "grok-voice-3.0", provider: "xai", transports: ["websocket"], available: true },
{ model: "gemini-3.8-live", provider: "google", transports: ["websocket"], available: true },
],
default: "grok-voice-3.0",
})
const none = yield* list({ openai: [credential("chatgpt", "oauth")] })
expect(none).toEqual({
models: [
{ model: "gpt-live-1", provider: "openai", transports: ["webrtc"], available: false },
{ model: "gpt-realtime-2.1", provider: "openai", transports: ["webrtc", "websocket"], available: false },
{ model: "grok-voice-3.0", provider: "xai", transports: ["websocket"], available: false },
{ model: "gemini-3.8-live", provider: "google", transports: ["websocket"], available: false },
],
})
expect("default" in none).toBe(false)
}),
)
+118
View File
@@ -5382,6 +5382,68 @@
}
}
},
"/api/experimental/voice/models": {
"get": {
"tags": ["voice"],
"operationId": "experimental.voice.models",
"parameters": [],
"security": [],
"responses": {
"200": {
"description": "VoiceModelsResponse",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/VoiceModelsResponse"
}
}
}
},
"400": {
"description": "InvalidRequestError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/InvalidRequestErrorEncoded"
}
}
}
},
"401": {
"description": "UnauthorizedError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/UnauthorizedErrorEncoded"
}
}
}
},
"404": {
"description": "LocationNotFoundError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/LocationNotFoundErrorEncoded"
}
}
}
},
"503": {
"description": "ServiceUnavailableError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ServiceUnavailableErrorEncoded"
}
}
}
}
},
"description": "List supported voice models, their transports, and whether the server has a usable API key for each provider.",
"summary": "List voice models"
}
},
"/api/experimental/voice/session": {
"post": {
"tags": ["voice"],
@@ -20536,6 +20598,58 @@
"required": ["_tag", "providerID", "message"],
"additionalProperties": false
},
"VoiceModel": {
"type": "object",
"properties": {
"model": {
"type": "string"
},
"provider": {
"type": "string"
},
"transports": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceTransportType"
}
},
"available": {
"type": "boolean"
}
},
"required": ["model", "provider", "transports", "available"],
"additionalProperties": false
},
"VoiceModelsResponse": {
"type": "object",
"properties": {
"data": {
"type": "object",
"properties": {
"models": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceModel"
}
},
"default": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
]
}
},
"required": ["models"],
"additionalProperties": false
}
},
"required": ["data"],
"additionalProperties": false
},
"VoiceSessionResponse": {
"type": "object",
"properties": {
@@ -20573,6 +20687,10 @@
}
]
},
"VoiceTransportType": {
"type": "string",
"enum": ["webrtc", "websocket"]
},
"VoiceWebRTCTransport": {
"type": "object",
"properties": {
+29
View File
@@ -2,6 +2,17 @@ import { Schema } from "effect"
import { HttpApiEndpoint, HttpApiGroup, OpenApi } from "effect/http-api"
import { InvalidRequestError, LocationNotFoundError, ServiceUnavailableError } from "../errors.js"
export const VoiceTransportType = Schema.Literals(["webrtc", "websocket"]).annotate({
identifier: "VoiceTransportType",
})
export const VoiceModel = Schema.Struct({
model: Schema.String,
provider: Schema.String,
transports: Schema.Array(VoiceTransportType),
available: Schema.Boolean,
}).annotate({ identifier: "VoiceModel" })
export const VoiceWebRTCTransportInput = Schema.Struct({
type: Schema.Literal("webrtc"),
sdp: Schema.String,
@@ -35,6 +46,24 @@ export const VoiceTransport = Schema.Union([
]).annotate({ identifier: "VoiceTransport" })
export const VoiceGroup = HttpApiGroup.make("server.voice")
.add(
HttpApiEndpoint.get("voice.models", "/api/experimental/voice/models", {
success: Schema.Struct({
data: Schema.Struct({
models: Schema.Array(VoiceModel),
default: Schema.optional(Schema.String),
}),
}).annotate({ identifier: "VoiceModelsResponse" }),
error: [ServiceUnavailableError, LocationNotFoundError],
}).annotateMerge(
OpenApi.annotations({
identifier: "experimental.voice.models",
summary: "List voice models",
description:
"List supported voice models, their transports, and whether the server has a usable API key for each provider.",
}),
),
)
.add(
HttpApiEndpoint.post("voice.session", "/api/experimental/voice/session", {
payload: Schema.Struct({
+34 -21
View File
@@ -14,26 +14,39 @@ export const VoiceHandler = HttpApiBuilder.group(Api, "server.voice", (handlers)
const global = yield* Global.Service
const locations = yield* LocationServiceMap.Service
const services = locations.get(Location.Ref.make({ directory: AbsolutePath.make(global.config) }))
return handlers.handle(
"voice.session",
Effect.fn("server.voice.session")(
function* (request) {
yield* awaitPlugins
const voice = yield* Voice.Service
const transport = yield* voice
.session(request.payload)
.pipe(
Effect.mapError((error) =>
error._tag === "Voice.InvalidRequestError"
? new InvalidRequestError({ message: error.message })
: new ServiceUnavailableError({ message: error.message, service: error.service }),
),
)
return { data: { transport } }
},
Effect.provide(services),
locationErrors,
),
)
return handlers
.handle(
"voice.models",
Effect.fn("server.voice.models")(
function* () {
yield* awaitPlugins
const voice = yield* Voice.Service
return { data: yield* voice.models() }
},
Effect.provide(services),
locationErrors,
),
)
.handle(
"voice.session",
Effect.fn("server.voice.session")(
function* (request) {
yield* awaitPlugins
const voice = yield* Voice.Service
const transport = yield* voice
.session(request.payload)
.pipe(
Effect.mapError((error) =>
error._tag === "Voice.InvalidRequestError"
? new InvalidRequestError({ message: error.message })
: new ServiceUnavailableError({ message: error.message, service: error.service }),
),
)
return { data: { transport } }
},
Effect.provide(services),
locationErrors,
),
)
}),
)
+27 -1
View File
@@ -13,6 +13,16 @@ const voice = makeLocationNode({
layer: Layer.succeed(
Voice.Service,
Voice.Service.of({
models: () =>
Effect.succeed({
models: [
{ model: "gpt-live-1", provider: "openai", transports: ["webrtc"], available: true },
{ model: "gpt-realtime-2.1", provider: "openai", transports: ["webrtc", "websocket"], available: true },
{ model: "grok-voice-3.0", provider: "xai", transports: ["websocket"], available: false },
{ model: "gemini-3.8-live", provider: "google", transports: ["websocket"], available: false },
],
default: "gpt-live-1",
}),
session: (input) => {
if (input.model === "invalid")
return Effect.fail(new Voice.InvalidRequestError({ message: "Unsupported voice model" }))
@@ -34,7 +44,7 @@ const voice = makeLocationNode({
deps: [],
})
it.live("routes POST /api/experimental/voice/session and maps errors", () =>
it.live("routes experimental voice models and session endpoints", () =>
Effect.gen(function* () {
const tmp = yield* Effect.acquireDisposable(Effect.promise(() => tmpdir("opencode-voice-endpoint-")))
const global = path.join(tmp.path, "global")
@@ -48,6 +58,22 @@ it.live("routes POST /api/experimental/voice/session and maps errors", () =>
},
{ overrides: [Voice.node.replace(voice)] },
)
const modelsRes = yield* Effect.promise(() =>
handler(new Request("http://opencode.local/api/experimental/voice/models")),
)
expect(modelsRes.status).toBe(200)
expect(yield* Effect.promise(() => modelsRes.json())).toEqual({
data: {
models: [
{ model: "gpt-live-1", provider: "openai", transports: ["webrtc"], available: true },
{ model: "gpt-realtime-2.1", provider: "openai", transports: ["webrtc", "websocket"], available: true },
{ model: "grok-voice-3.0", provider: "xai", transports: ["websocket"], available: false },
{ model: "gemini-3.8-live", provider: "google", transports: ["websocket"], available: false },
],
default: "gpt-live-1",
},
})
const post = (body: unknown) =>
Effect.promise(() =>
handler(
+118
View File
@@ -5382,6 +5382,68 @@
}
}
},
"/api/experimental/voice/models": {
"get": {
"tags": ["voice"],
"operationId": "experimental.voice.models",
"parameters": [],
"security": [],
"responses": {
"200": {
"description": "VoiceModelsResponse",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/VoiceModelsResponse"
}
}
}
},
"400": {
"description": "InvalidRequestError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/InvalidRequestErrorEncoded"
}
}
}
},
"401": {
"description": "UnauthorizedError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/UnauthorizedErrorEncoded"
}
}
}
},
"404": {
"description": "LocationNotFoundError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/LocationNotFoundErrorEncoded"
}
}
}
},
"503": {
"description": "ServiceUnavailableError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ServiceUnavailableErrorEncoded"
}
}
}
}
},
"description": "List supported voice models, their transports, and whether the server has a usable API key for each provider.",
"summary": "List voice models"
}
},
"/api/experimental/voice/session": {
"post": {
"tags": ["voice"],
@@ -20536,6 +20598,58 @@
"required": ["_tag", "providerID", "message"],
"additionalProperties": false
},
"VoiceModel": {
"type": "object",
"properties": {
"model": {
"type": "string"
},
"provider": {
"type": "string"
},
"transports": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceTransportType"
}
},
"available": {
"type": "boolean"
}
},
"required": ["model", "provider", "transports", "available"],
"additionalProperties": false
},
"VoiceModelsResponse": {
"type": "object",
"properties": {
"data": {
"type": "object",
"properties": {
"models": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceModel"
}
},
"default": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
]
}
},
"required": ["models"],
"additionalProperties": false
}
},
"required": ["data"],
"additionalProperties": false
},
"VoiceSessionResponse": {
"type": "object",
"properties": {
@@ -20573,6 +20687,10 @@
}
]
},
"VoiceTransportType": {
"type": "string",
"enum": ["webrtc", "websocket"]
},
"VoiceWebRTCTransport": {
"type": "object",
"properties": {
+118
View File
@@ -5382,6 +5382,68 @@
}
}
},
"/api/experimental/voice/models": {
"get": {
"tags": ["voice"],
"operationId": "experimental.voice.models",
"parameters": [],
"security": [],
"responses": {
"200": {
"description": "VoiceModelsResponse",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/VoiceModelsResponse"
}
}
}
},
"400": {
"description": "InvalidRequestError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/InvalidRequestErrorEncoded"
}
}
}
},
"401": {
"description": "UnauthorizedError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/UnauthorizedErrorEncoded"
}
}
}
},
"404": {
"description": "LocationNotFoundError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/LocationNotFoundErrorEncoded"
}
}
}
},
"503": {
"description": "ServiceUnavailableError",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ServiceUnavailableErrorEncoded"
}
}
}
}
},
"description": "List supported voice models, their transports, and whether the server has a usable API key for each provider.",
"summary": "List voice models"
}
},
"/api/experimental/voice/session": {
"post": {
"tags": ["voice"],
@@ -20536,6 +20598,58 @@
"required": ["_tag", "providerID", "message"],
"additionalProperties": false
},
"VoiceModel": {
"type": "object",
"properties": {
"model": {
"type": "string"
},
"provider": {
"type": "string"
},
"transports": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceTransportType"
}
},
"available": {
"type": "boolean"
}
},
"required": ["model", "provider", "transports", "available"],
"additionalProperties": false
},
"VoiceModelsResponse": {
"type": "object",
"properties": {
"data": {
"type": "object",
"properties": {
"models": {
"type": "array",
"items": {
"$ref": "#/components/schemas/VoiceModel"
}
},
"default": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
]
}
},
"required": ["models"],
"additionalProperties": false
}
},
"required": ["data"],
"additionalProperties": false
},
"VoiceSessionResponse": {
"type": "object",
"properties": {
@@ -20573,6 +20687,10 @@
}
]
},
"VoiceTransportType": {
"type": "string",
"enum": ["webrtc", "websocket"]
},
"VoiceWebRTCTransport": {
"type": "object",
"properties": {