synthesize() - RunAnywhere Documentation

Three methods, depending on whether you want the samples or just want it spoken.

import RunAnywhereONNX
ONNX.register()

// Get the audio back
let audio = try await RunAnywhere.tts.synthesize("Hello from RunAnywhere")

// Play it through the device
let handle = try await RunAnywhere.tts.speak("Hello from RunAnywhere")
await handle.waitForPlayout()

Signatures

func synthesize(_ text: String, options: TtsOptions? = nil) async throws -> Audio
func synthesizeStream(_ text: String, options: TtsOptions? = nil) async throws
    -> AsyncThrowingStream<AudioChunk, Error>
func speak(_ text: String, options: TtsOptions? = nil) async throws -> SpeechHandle
func stop() async
func voices() async throws -> [Voice]

synthesize gives you an Audio value to save or feed elsewhere. speak plays it and hands back a SpeechHandle. synthesizeStream is covered in Streaming TTS.

SpeechHandle

let handle = try await RunAnywhere.tts.speak("A long passage of text…")

// Wait for it to finish
await handle.waitForPlayout()

// Or cut it off
await handle.interrupt()

if handle.interrupted { print("stopped early") }
if let error = handle.error { print(error) }

RunAnywhere.tts.stop() stops whatever is currently speaking, without a handle.

Options

let audio = try await RunAnywhere.tts.synthesize(
    "Hello",
    options: TtsOptions(voice: "en_US-amy-medium", speed: 1.1)
)

Field	Type	Default
`voice`	`String?`	none, model default
`language`	`String`	proto default
`speed`	`Float`	proto default
`pitch`	`Float`	proto default
`format`	`RAAudioFormat`	proto default
`sampleRate`	`Int`	proto default

See Voices for choosing one.

Saving to a file

let audio = try await RunAnywhere.tts.synthesize("Save me")
try audio.samples.write(to: url)

Audio exposes duration alongside its samples.