@rimelabs/sdk 0.1.0-alpha.2 → 0.1.0-alpha.20

package/dist/tts/client.d.ts

@@ -0,0 +1,53 @@
+import { AudioFormat } from "./audio.js";
+import { type PronunciationEntry } from "./lexicon.js";
+import { AudioStream, type TextSource } from "./stream.js";
+import { Credentials } from "../auth.js";
+import { policy } from "./policy.js";
+export interface SynthesisOptions {
+    voice?: string | null;
+    language?: string;
+    audioFormat?: AudioFormat | null;
+    /** Requested output rate: 8000, 16000, or 24000 Hz. Defaults to 24000 for PCM/WAV/MP3. */
+    sampleRate?: number | null;
+    /** Duration multiplier from 0.4 to 2.5. Below 1 speeds speech up; above 1 slows it down. */
+    timeScaleFactor?: number | null;
+    /** Request final word timestamps. Supported only with model="mistv3". */
+    timestamps?: boolean;
+    /** Request-wide pronunciation overrides; requires a phoneme-capable Coda deployment. */
+    customLexicon?: readonly PronunciationEntry[];
+    timeout?: number | null;
+}
+export interface DiscoveryOptions {
+    timeout?: number | null;
+}
+export interface VoiceListOptions extends DiscoveryOptions {
+    language?: string | null;
+}
+export declare class TTS {
+    private readonly credentials;
+    private readonly deployment;
+    private readonly defaultTimeout;
+    private readonly checkOpen;
+    private client;
+    private streams;
+    private discoveryOperations;
+    readonly voices: {
+        list: (options?: VoiceListOptions) => Promise<string[]>;
+    };
+    readonly languages: {
+        list: (options?: DiscoveryOptions) => Promise<string[]>;
+    };
+    constructor(credentials: Credentials, deployment: Readonly<typeof policy>, defaultTimeout: number | null, checkOpen: () => void);
+    private prepare;
+    /** Synthesize complete text and return an audio stream. Do not await this call. */
+    synthesize(text: string, options?: SynthesisOptions): AudioStream;
+    /** Await complete audio in memory, finalizing WAV lengths. Never returns partial audio.
+     * The fixed output limit is 64 MiB (67,108,864 bytes); exceeding it raises
+     * RimeResourceLimitError. Use streaming synthesis for larger outputs or timestamps.
+     */
+    synthesizeToBuffer(text: string, options?: SynthesisOptions): Promise<Buffer>;
+    /** Stream text chunks into synthesis and return an audio stream. Do not await this call. */
+    stream(text: TextSource, options?: SynthesisOptions): AudioStream;
+    private discover;
+    close(): Promise<void>;
+}

package/dist/tts/client.js

@@ -0,0 +1,151 @@
+import { AudioFormat, resolveAudio } from "./audio.js";
+import { maxBufferBytes, finalizeWav } from "./buffer.js";
+import { snapshot } from "./lexicon.js";
+import { AudioStream, constructionKey } from "./stream.js";
+import { policy, timeout, nonempty } from "./policy.js";
+import { abortable } from "../cancellation.js";
+import { transport, connect, metadata, discover, } from "./transport.js";
+import { RimeInputError, RimeResourceLimitError, RimeCancelledError, RimeTimeoutError, RimeUnavailableError, } from "../errors.js";
+export class TTS {
+    credentials;
+    deployment;
+    defaultTimeout;
+    checkOpen;
+    client = null;
+    streams = new Set();
+    discoveryOperations = new Map();
+    voices;
+    languages;
+    constructor(credentials, deployment, defaultTimeout, checkOpen) {
+        this.credentials = credentials;
+        this.deployment = deployment;
+        this.defaultTimeout = defaultTimeout;
+        this.checkOpen = checkOpen;
+        this.voices = {
+            list: async (options = {}) => {
+                if (options.language !== undefined && options.language !== null)
+                    nonempty(options.language, "language");
+                return this.discover("voices", options.language, options.timeout);
+            },
+        };
+        this.languages = {
+            list: (options = {}) => this.discover("languages", undefined, options.timeout),
+        };
+    }
+    async prepare(signal) {
+        this.checkOpen();
+        const auth = await this.credentials.metadata(signal);
+        this.checkOpen();
+        this.client ??= transport.makeClient(this.deployment.target);
+        await connect(this.client, signal);
+        return { client: this.client, metadata: metadata(auth) };
+    }
+    /** Synthesize complete text and return an audio stream. Do not await this call. */
+    synthesize(text, options = {}) {
+        this.checkOpen();
+        if (typeof text !== "string")
+            throw new RimeInputError("text must be a string; use stream() for an async text source");
+        if (!text.trim())
+            throw new RimeInputError("Text must contain non-whitespace characters");
+        return this.stream((async function* () {
+            yield text;
+        })(), options);
+    }
+    /** Await complete audio in memory, finalizing WAV lengths. Never returns partial audio.
+     * The fixed output limit is 64 MiB (67,108,864 bytes); exceeding it raises
+     * RimeResourceLimitError. Use streaming synthesis for larger outputs or timestamps.
+     */
+    async synthesizeToBuffer(text, options = {}) {
+        const audio = this.synthesize(text, options);
+        const chunks = [];
+        let size = 0;
+        try {
+            for await (const chunk of audio) {
+                if (chunk.byteLength > maxBufferBytes - size)
+                    throw new RimeResourceLimitError("Audio exceeds the 64 MiB buffer limit", audio.requestId);
+                chunks.push(chunk);
+                size += chunk.byteLength;
+            }
+            const data = Buffer.concat(chunks, size);
+            if (audio.format === AudioFormat.WAV)
+                finalizeWav(data, audio.sampleRate, audio.requestId);
+            return data;
+        }
+        finally {
+            await audio.cancel();
+        }
+    }
+    /** Stream text chunks into synthesis and return an audio stream. Do not await this call. */
+    stream(text, options = {}) {
+        this.checkOpen();
+        if (typeof text === "string" ||
+            !text ||
+            typeof text[Symbol.asyncIterator] !== "function")
+            throw new RimeInputError("text must be an async iterable of strings; use synthesize() for a string");
+        const lexicon = snapshot(options.customLexicon);
+        const timestamps = options.timestamps === undefined ? false : options.timestamps;
+        if (typeof timestamps !== "boolean")
+            throw new RimeInputError("timestamps must be a boolean");
+        if (timestamps && this.deployment.model !== "mistv3")
+            throw new RimeInputError("Word timestamps are supported only with model='mistv3'");
+        const voice = options.voice === undefined || options.voice === null
+            ? this.deployment.defaultVoice
+            : nonempty(options.voice, "voice");
+        const language = nonempty(options.language ?? "en", "language");
+        const { format, sampleRate } = resolveAudio(options.audioFormat, options.sampleRate, options.timeScaleFactor);
+        const audio = new AudioStream(constructionKey, {
+            prepare: (s) => this.prepare(s),
+            checkOpen: () => this.checkOpen(),
+            forget: (s) => this.streams.delete(s),
+        }, text, voice, language, format, timeout(options.timeout, this.defaultTimeout), timestamps, lexicon, sampleRate, options.timeScaleFactor ?? undefined);
+        this.streams.add(audio);
+        return audio;
+    }
+    async discover(kind, language, override) {
+        this.checkOpen();
+        const budget = Math.min(timeout(override, this.defaultTimeout) ?? Infinity, policy.discoveryTimeout);
+        const controller = new AbortController();
+        let requestId = null;
+        const timer = setTimeout(() => controller.abort(new RimeTimeoutError("Discovery deadline expired")), budget * 1000);
+        const operation = (async () => {
+            try {
+                const prepared = await this.prepare(controller.signal);
+                for (let attempt = 0;; attempt++) {
+                    requestId = null;
+                    try {
+                        return await discover(prepared, kind, language, controller.signal);
+                    }
+                    catch (error) {
+                        if (!(error instanceof RimeUnavailableError) || attempt === 2)
+                            throw error;
+                        requestId = error.requestId;
+                        await abortable(new Promise((resolve) => setTimeout(resolve, 50 * 2 ** attempt)), controller.signal);
+                    }
+                }
+            }
+            catch (error) {
+                if (error instanceof RimeTimeoutError && requestId !== null)
+                    throw new RimeTimeoutError(error.message, requestId, {
+                        cause: error.cause,
+                    });
+                throw error;
+            }
+            finally {
+                clearTimeout(timer);
+                this.discoveryOperations.delete(controller);
+            }
+        })();
+        this.discoveryOperations.set(controller, operation);
+        return operation;
+    }
+    async close() {
+        const discoveries = [...this.discoveryOperations.values()];
+        for (const controller of this.discoveryOperations.keys())
+            controller.abort(new RimeCancelledError("Client closed during discovery"));
+        await Promise.all([
+            ...[...this.streams].map((stream) => stream.cancel()),
+            Promise.allSettled(discoveries),
+        ]);
+        this.client?.close();
+    }
+}

package/dist/tts/lexicon.d.ts

@@ -0,0 +1,7 @@
+/** A word or phrase and its space-separated X-SAMPA pronunciation. */
+export interface PronunciationEntry {
+    readonly spelling: string;
+    readonly pronunciation: string;
+}
+/** Copy request options now; model and linguistic validation belong to the service. */
+export declare function snapshot(entries?: readonly PronunciationEntry[]): readonly PronunciationEntry[];

package/dist/tts/lexicon.js

@@ -0,0 +1,23 @@
+import { RimeInputError } from "../errors.js";
+// Unicode mode treats valid surrogate pairs as one code point outside this range.
+const unpairedSurrogate = /[\uD800-\uDFFF]/u;
+/** Copy request options now; model and linguistic validation belong to the service. */
+export function snapshot(entries = []) {
+    if (!Array.isArray(entries))
+        throw new RimeInputError("customLexicon must be an array of pronunciation entries");
+    // Array.from also visits holes, which are invalid entries.
+    return Object.freeze(Array.from(entries, (entry) => {
+        if (!entry ||
+            typeof entry !== "object" ||
+            typeof entry.spelling !== "string" ||
+            typeof entry.pronunciation !== "string")
+            throw new RimeInputError("Lexicon entries must have string spelling and pronunciation fields");
+        if (unpairedSurrogate.test(entry.spelling) ||
+            unpairedSurrogate.test(entry.pronunciation))
+            throw new RimeInputError("Lexicon spelling and pronunciation must contain valid Unicode");
+        return Object.freeze({
+            spelling: entry.spelling,
+            pronunciation: entry.pronunciation,
+        });
+    }));
+}

package/dist/{policy.d.ts → tts/policy.d.ts}

@@ -1,9 +1,10 @@
+export { timeout } from "../validation.js";
 export declare const policy: {
+    model: string;
     target: string;
     audience: string;
     defaultVoice: string;
     exchangeUrl: string;
-    sentenceBytes: number;
     sourceChars: number;
     outputBytes: number;
     outputChunkBytes: number;
@@ -16,6 +17,4 @@ export declare const policy: {
     cleanupTimeout: number;
 };
 export declare function resolve(model: string, endpoint?: string | null): Readonly<typeof policy>;
-export declare function timeout(value: unknown, inherited?: number | null): number | null;
 export declare function nonempty(value: unknown, name: string): string;
-export declare function abortable<T>(promise: PromiseLike<T>, signal: AbortSignal): Promise<T>;

package/dist/tts/policy.js

@@ -0,0 +1,41 @@
+import { endpointAddress } from "../validation.js";
+export { timeout } from "../validation.js";
+import { RimeInputError } from "../errors.js";
+const codaHostname = "coda.api.rime.ai";
+export const policy = {
+    model: "coda",
+    target: `${codaHostname}:443`,
+    audience: codaHostname,
+    defaultVoice: "clementine",
+    exchangeUrl: "https://themis.api.rime.ai/v1/token",
+    sourceChars: 1024,
+    outputBytes: 96000,
+    outputChunkBytes: 9600,
+    receiveBytes: 4194304,
+    authTimeout: 10,
+    connectionTimeout: 10,
+    firstAudioTimeout: 30,
+    progressTimeout: 60,
+    discoveryTimeout: 10,
+    cleanupTimeout: 2,
+};
+export function resolve(model, endpoint) {
+    const deployment = { ...policy, model };
+    if (model === "mistv3") {
+        const hostname = "mist.api.rime.ai";
+        deployment.target = `${hostname}:443`;
+        deployment.audience = hostname;
+        deployment.defaultVoice = "astra";
+    }
+    else if (model !== "coda") {
+        throw new RimeInputError("model must be 'coda' or 'mistv3'");
+    }
+    if (endpoint === undefined || endpoint === null)
+        return deployment;
+    return { ...deployment, ...endpointAddress(endpoint) };
+}
+export function nonempty(value, name) {
+    if (typeof value !== "string" || !value.trim())
+        throw new RimeInputError(`${name} must be a non-empty string`);
+    return value;
+}

package/dist/{stream.d.ts → tts/stream.d.ts}

@@ -1,11 +1,13 @@
 import { AudioFormat } from "./audio.js";
+import type { TimestampResult } from "./timestamps.js";
+import type { PronunciationEntry } from "./lexicon.js";
 import { type PreparedConnection } from "./transport.js";
 export interface StreamOwner {
     prepare(signal: AbortSignal): Promise<PreparedConnection>;
     checkOpen(): void;
     forget(stream: AudioStream): void;
 }
-export type TextSource = string | AsyncIterable<string>;
+export type TextSource = AsyncIterable<string>;
 export declare const constructionKey: unique symbol;
 export declare class AudioStream implements AsyncIterableIterator<Uint8Array> {
     private owner;
@@ -14,6 +16,10 @@ export declare class AudioStream implements AsyncIterableIterator<Uint8Array> {
     private language;
     private formatValue;
     private timeout;
+    private timestampsRequested;
+    private customLexicon;
+    private sampleRateValue;
+    private timeScaleFactor?;
     private readonly controller;
     private readonly queue;
     private worker;
@@ -31,9 +37,16 @@ export declare class AudioStream implements AsyncIterableIterator<Uint8Array> {
     private readonly operationId;
     private bytesReceived;
     private bytesDelivered;
-    constructor(key: symbol, owner: StreamOwner, source: TextSource | null, voice: string, language: string, formatValue: AudioFormat, timeout: number | null);
+    constructor(key: symbol, owner: StreamOwner, source: TextSource | null, voice: string, language: string, formatValue: AudioFormat, timeout: number | null, timestampsRequested?: boolean, customLexicon?: readonly PronunciationEntry[], sampleRateValue?: number, timeScaleFactor?: number | undefined);
     get format(): AudioFormat;
+    /** Requested output rate. MP3 headers describe the encoded rate. */
+    get sampleRate(): number;
     get requestId(): string | null;
+    /** Read requested timestamps after consuming all audio; never drains audio.
+     * Calling before iteration completes rejects with RimeInputError. Synthesis
+     * errors reject here too; alignment failures are returned in status.
+     */
+    timestamps(): Promise<TimestampResult>;
     private start;
     private fail;
     private produce;

package/dist/{stream.js → tts/stream.js}

@@ -1,12 +1,12 @@
 import { randomUUID } from "node:crypto";
 import { debuglog } from "node:util";
 const log = debuglog("rime-sdk");
-import { Converter } from "./audio.js";
-import { policy, abortable } from "./policy.js";
+import { AudioFormat, Converter, wireFormat } from "./audio.js";
+import { policy } from "./policy.js";
+import { abortable } from "../cancellation.js";
 import { ByteQueue } from "./queue.js";
-import { SentenceBuffer, ready } from "./sentences.js";
 import { SynthesisCall } from "./transport.js";
-import { RimeError, RimeInputError, RimeCancelledError, RimeTimeoutError, RimeStreamError, } from "./errors.js";
+import { RimeError, RimeInputError, RimeCancelledError, RimeTimeoutError, RimeStreamError, } from "../errors.js";
 export const constructionKey = Symbol("private stream constructor");
 export class AudioStream {
     owner;
@@ -15,6 +15,10 @@ export class AudioStream {
     language;
     formatValue;
     timeout;
+    timestampsRequested;
+    customLexicon;
+    sampleRateValue;
+    timeScaleFactor;
     controller = new AbortController();
     queue = new ByteQueue(policy.outputBytes, policy.outputChunkBytes);
     worker = null;
@@ -32,22 +36,43 @@ export class AudioStream {
     operationId = randomUUID();
     bytesReceived = 0;
     bytesDelivered = 0;
-    constructor(key, owner, source, voice, language, formatValue, timeout) {
+    constructor(key, owner, source, voice, language, formatValue, timeout, timestampsRequested = false, customLexicon = [], sampleRateValue = 24000, timeScaleFactor) {
         this.owner = owner;
         this.source = source;
         this.voice = voice;
         this.language = language;
         this.formatValue = formatValue;
         this.timeout = timeout;
+        this.timestampsRequested = timestampsRequested;
+        this.customLexicon = customLexicon;
+        this.sampleRateValue = sampleRateValue;
+        this.timeScaleFactor = timeScaleFactor;
         if (key !== constructionKey)
-            throw new TypeError("AudioStream is returned by client.tts.stream()");
+            throw new TypeError("AudioStream is returned by client.tts.synthesize() or client.tts.stream()");
     }
     get format() {
         return this.formatValue;
     }
+    /** Requested output rate. MP3 headers describe the encoded rate. */
+    get sampleRate() {
+        return this.sampleRateValue;
+    }
     get requestId() {
         return this.call?.requestId ?? null;
     }
+    /** Read requested timestamps after consuming all audio; never drains audio.
+     * Calling before iteration completes rejects with RimeInputError. Synthesis
+     * errors reject here too; alignment failures are returned in status.
+     */
+    async timestamps() {
+        if (!this.timestampsRequested)
+            throw new RimeInputError("Enable timestamps: true when creating the stream");
+        if (this.failure)
+            throw this.failure;
+        if (!this.finished)
+            throw new RimeInputError("Consume all audio before reading timestamps");
+        return this.call.timestampResult();
+    }
     start() {
         if (this.failure)
             throw this.failure;
@@ -95,15 +120,11 @@ export class AudioStream {
     async produce() {
         const source = this.source;
         let iterator;
-        const buffer = new SentenceBuffer(policy.sentenceBytes);
+        let highSurrogate = "";
         let meaningful = false;
         try {
             try {
-                iterator = (typeof source === "string"
-                    ? (async function* () {
-                        yield source;
-                    })()
-                    : source)[Symbol.asyncIterator]();
+                iterator = source[Symbol.asyncIterator]();
             }
             catch (error) {
                 if (this.controller.signal.aborted)
@@ -130,35 +151,39 @@ export class AudioStream {
                 }
                 if (result.done)
                     break;
-                const chunk = result.value;
-                if (typeof chunk !== "string")
+                if (typeof result.value !== "string")
                     throw new RimeInputError("The text source must yield strings");
+                let chunk = highSurrogate + result.value;
+                highSurrogate = "";
+                // A JavaScript source may split one Unicode character across yields.
+                if (/[\uD800-\uDBFF]$/.test(chunk)) {
+                    highSurrogate = chunk.slice(-1);
+                    chunk = chunk.slice(0, -1);
+                }
                 if (chunk.trim())
                     meaningful = true;
-                // Never split a surrogate pair when limiting native detector input size.
+                // Bound wire messages without splitting a Unicode character.
                 for (let offset = 0; offset < chunk.length;) {
                     let end = Math.min(chunk.length, offset + policy.sourceChars);
                     if (end < chunk.length && /[\uD800-\uDBFF]/.test(chunk[end - 1]))
                         end++;
-                    for (const sentence of buffer.feed(chunk.slice(offset, end))) {
-                        if (!this.submitted)
-                            this.progressAt = performance.now();
-                        this.submitted = true;
-                        await this.call.write(sentence);
-                    }
+                    const part = chunk.slice(offset, end);
+                    if (/[\uD800-\uDFFF]/u.test(part))
+                        throw new RimeInputError("Text must contain valid Unicode");
+                    if (!this.submitted)
+                        this.progressAt = performance.now();
+                    this.submitted = true;
+                    await this.call.write(part);
+                    this.progressAt = performance.now();
                     offset = end;
                 }
             }
+            if (highSurrogate)
+                throw new RimeInputError("Text must contain valid Unicode");
             if (!meaningful)
                 throw new RimeInputError("The text source contained no meaningful text");
-            for (const sentence of buffer.feed("", true)) {
-                if (!this.submitted)
-                    this.progressAt = performance.now();
-                this.submitted = true;
-                await this.call.write(sentence);
-            }
-            this.call.finishInput();
             this.inputDone = true;
+            this.call.finishInput();
         }
         catch (error) {
             if (error instanceof RimeError || this.controller.signal.aborted)
@@ -199,10 +224,9 @@ export class AudioStream {
     async run() {
         let tasks = [];
         try {
-            await abortable(ready, this.controller.signal);
             const prepared = await this.owner.prepare(this.controller.signal);
             this.call = new SynthesisCall(prepared, this.controller.signal);
-            await this.call.start(this.voice, this.language);
+            await this.call.start(this.voice, this.language, this.timestampsRequested, this.customLexicon, wireFormat(this.format), this.format === AudioFormat.MULAW_8000 ? 24000 : this.sampleRate, this.timeScaleFactor);
             tasks = [this.produce(), this.read()];
             await Promise.all(tasks);
             this.queue.finish();

package/dist/tts/timestamps.d.ts

@@ -0,0 +1,24 @@
+import type { SynthesisResponseTrailer } from "@rimelabs/api";
+/** A spoken, normalized word with start/end seconds from synthesis start. */
+export interface WordTimestamp {
+    readonly text: string;
+    readonly start: number;
+    readonly end: number;
+}
+/** Alignment's google.rpc.Code (0 means OK) and service explanation. */
+export interface TimestampStatus {
+    readonly code: number;
+    readonly message: string;
+}
+/** Final alignment status and words. Non-OK statuses carry no spans. */
+export interface TimestampResult {
+    readonly status: TimestampStatus;
+    readonly spans: readonly WordTimestamp[];
+}
+export declare class TimestampTrailer {
+    seen: boolean;
+    invalid: boolean;
+    private value;
+    accept(trailer: SynthesisResponseTrailer): void;
+    result(requestId: string | null): TimestampResult;
+}

package/dist/tts/timestamps.js

@@ -0,0 +1,46 @@
+import { RimeStreamError } from "../errors.js";
+export class TimestampTrailer {
+    seen = false;
+    invalid = false;
+    value;
+    accept(trailer) {
+        if (this.seen)
+            this.invalid = true;
+        else {
+            this.seen = true;
+            this.value = trailer.timestamps;
+        }
+    }
+    result(requestId) {
+        const malformed = () => new RimeStreamError("Invalid timestamp trailer", requestId);
+        if (this.invalid)
+            throw malformed();
+        const value = this.value;
+        if (!value)
+            throw new RimeStreamError("The service did not return requested timestamps", requestId);
+        if (!value.status || (value.status.code !== 0 && value.spans.length))
+            throw malformed();
+        const seconds = (duration) => {
+            if (!duration ||
+                duration.seconds < 0n ||
+                duration.seconds > 315576000000n ||
+                duration.nanos < 0 ||
+                duration.nanos >= 1000000000)
+                throw malformed();
+            return Number(duration.seconds) + duration.nanos / 1e9;
+        };
+        const spans = value.spans.map((span) => {
+            const start = seconds(span.start), end = seconds(span.end);
+            if (!span.text.trim() || end < start)
+                throw malformed();
+            return Object.freeze({ text: span.text, start, end });
+        });
+        return Object.freeze({
+            status: Object.freeze({
+                code: value.status.code,
+                message: value.status.message,
+            }),
+            spans: Object.freeze(spans),
+        });
+    }
+}

package/dist/{transport.d.ts → tts/transport.d.ts}

@@ -1,33 +1,30 @@
+export { metadata } from "../grpc.js";
+export { rpcError } from "../grpc.js";
 import * as grpc from "@grpc/grpc-js";
-import * as errors from "./errors.js";
+import type { PronunciationEntry } from "./lexicon.js";
 export type Connection = grpc.Client;
 export interface PreparedConnection {
     client: Connection;
     metadata: grpc.Metadata;
 }
 export type DiscoveryKind = "voices" | "languages";
-export declare function rpcError(code: number, requestId?: string | null): errors.RimeError;
 export declare const transport: {
     makeClient: (target: string) => grpc.Client;
 };
 export declare function connect(client: grpc.Client, signal: AbortSignal): Promise<void>;
-export declare function metadata(value: string): grpc.Metadata;
 export declare function discover(prepared: PreparedConnection, kind: DiscoveryKind, language: string | null | undefined, signal: AbortSignal): Promise<string[]>;
 export declare class SynthesisCall {
-    private signal;
-    private readonly call;
-    private readonly metadataPromise;
-    private readonly statusPromise;
-    private status;
-    private id;
+    private audioFormat;
+    private timestampsRequested;
+    private readonly timestampTrailer;
+    private readonly rpc;
     constructor(prepared: PreparedConnection, signal: AbortSignal);
     get requestId(): string | null;
     get done(): boolean;
     cancel(): void;
-    private error;
-    private writeMessage;
-    start(voice: string, language: string): Promise<void>;
-    write(sentence: string): Promise<void>;
+    start(voice: string, language: string, timestamps?: boolean, customLexicon?: readonly PronunciationEntry[], audioFormat?: string, sampleRate?: number, timeScaleFactor?: number): Promise<void>;
+    timestampResult(): import("./timestamps.js").TimestampResult;
+    write(text: string): Promise<void>;
     finishInput(): void;
     private checkFormat;
     audio(): AsyncGenerator<Uint8Array>;