@tanstack/ai
Version:
Type-safe TypeScript AI SDK for streaming chat, tool calling, agents, structured outputs, and multimodal generation.
134 lines (119 loc) • 4.09 kB
text/typescript
import type {
ListVoicesOptions,
ListVoicesResult,
TTSOptions,
TTSResult,
} from '../../types'
/**
* What a TTS adapter can do beyond a single voice reading a single string.
*
* Declared statically so `generateSpeech()` can reject an unsupported request
* before it reaches the provider, instead of surfacing a provider 422.
*/
export interface TTSCapabilities {
/**
* Maximum number of distinct voices accepted across `turns`
* (ElevenLabs 10, Gemini 2). Omit it when the adapter has no dialogue
* endpoint — then `turns` is rejected outright.
*/
maxSpeakers?: number
/** Set when the adapter can honour `timestamps: true`. */
timestamps?: boolean
}
/**
* Configuration for TTS adapter instances
*/
export interface TTSAdapterConfig {
apiKey?: string
baseUrl?: string
timeout?: number
maxRetries?: number
headers?: Record<string, string>
}
/**
* TTS adapter interface with pre-resolved generics.
*
* An adapter is created by a provider function: `provider('model')` → `adapter`
* All type resolution happens at the provider call site, not in this interface.
*
* Generic parameters:
* - TModel: The specific model name (e.g., 'tts-1')
* - TProviderOptions: Provider-specific options (already resolved)
*/
export interface TTSAdapter<
TModel extends string = string,
TProviderOptions extends object = Record<string, unknown>,
> {
/** Discriminator for adapter kind - used to determine API shape */
readonly kind: 'tts'
/** Adapter name identifier */
readonly name: string
/** The model this adapter is configured for */
readonly model: TModel
/**
* Optional static capability declaration. Absent means "single voice, no
* timestamps" — the contract every adapter had before dialogue existed.
*/
readonly capabilities?: TTSCapabilities
/**
* @internal Type-only properties for inference. Not assigned at runtime.
*/
'~types': {
providerOptions: TProviderOptions
}
/**
* Generate speech from text
*/
generateSpeech: (options: TTSOptions<TProviderOptions>) => Promise<TTSResult>
/**
* List the voices this account can use.
*
* Optional, because only some providers have a catalog worth querying at
* runtime. A provider whose voices are a fixed list known at build time
* publishes that list from its own package instead (`GeminiTTSVoices`, or
* the `OpenAITTSVoice` union), which is strictly better than a network
* call. Implement this only when the catalog is per-account and can change,
* which is the case wherever `generateVoice()` can add to it.
*/
listVoices?: (options?: ListVoicesOptions) => Promise<ListVoicesResult>
}
/**
* A TTSAdapter with any/unknown type parameters.
* Useful as a constraint in generic functions and interfaces.
*/
export type AnyTTSAdapter = TTSAdapter<any, any>
/**
* Abstract base class for text-to-speech adapters.
* Extend this class to implement a TTS adapter for a specific provider.
*
* Generic parameters match TTSAdapter - all pre-resolved by the provider function.
*/
export abstract class BaseTTSAdapter<
TModel extends string = string,
TProviderOptions extends object = Record<string, unknown>,
> implements TTSAdapter<TModel, TProviderOptions> {
readonly kind = 'tts' as const
abstract readonly name: string
readonly model: TModel
declare readonly capabilities?: TTSCapabilities
// Type-only property - never assigned at runtime
declare '~types': {
providerOptions: TProviderOptions
}
protected config: TTSAdapterConfig
constructor(model: TModel, config: TTSAdapterConfig = {}) {
this.config = config
this.model = model
}
abstract generateSpeech(
options: TTSOptions<TProviderOptions>,
): Promise<TTSResult>
/**
* Not abstract: a provider with a fixed voice list has nothing to query and
* should not be forced to write a stub.
*/
listVoices?(options?: ListVoicesOptions): Promise<ListVoicesResult>
protected generateId(): string {
return `${this.name}-${Date.now()}-${Math.random().toString(36).substring(7)}`
}
}