Types of voices

Text-to-Speech generates audio with natural, human-like quality, which creates speech that sounds like a real person. To start, specify a voice when sending a synthesis request.

Text-to-Speech offers a variety of voices based on language, gender, and accent. Some languages have multiple options. For a full list, check the Supported Voices page. To select a voice, use the VoiceSelectionParams field in your API request. Refer to the Quickstarts for instructions on making a synthesize request.

Overview

Voice Type		Intended for	Launch stage	Controllability	Streaming
Journey		Conversational Agents	Preview	-	Yes
Studio	Two speakers group	Media - Discussions and Interviews	Experimental	-	-
	One speaker person	Media - Narration	GA	SSML	-
Neural2		General purpose	GA	SSML	-
Standard		Cost efficient	GA	SSML	-

Pricing Details

Journey voices

Journey Voices, powered by the AudioLM engine, lets you create more engaging and empathetic speech for conversational applications. Through text streaming, Journey Voices produces low-latency real-time communication and supports the languages listed in the table of supported voices.

Note: Journey Voices, which is powered by LLMs, can hallucinate by dropping words or digits during synthesis.

Chat experiences

Voice: en-US-Journey-F

Other examples

Virtual assistants

Voice: en-US-Journey-D

Customer service chatbots

Voice: en-US-Journey-F

Interactive education applications

Voice: en-US-Journey-O

Sales and pitches

Voice: en-US-Journey-D

Storytime

Voice: en-US-Journey-F

Note: Journey Voices doesn't support SSML input, speaking rate and pitch audio parameters, and the A-Law audio encoding.

Note: Journey voices is available in the global, eu, and us endpoints but is out of scope for regionalization and data residency.

Studio multispeaker voices

Create discussions and interviews with the new multispeaker studio voices, based on the same technology behind Journey voices.

Studio voices

Studio voices are designed for news reading and broadcast content.

Example 1. The en-US-Studio-O voice reading the Great Gatsby.

Note: Studio voices support SSML, except for the following tags: <mark>, <emphasis>, <prosody pitch>, and <lang>.

Note: Check the table of supported voices for availability of Studio voices in specific languages.

Neural2 voices

The Text-to-Speech API provides a voice tier called Neural2. Neural2 voices are based on the same technology used to create a Custom Voice. Neural2 allows anyone to use Custom Voice technology without training their own custom voice. They're available in global and single region endpoints.

Note: Check the table of supported voices for availability of Neural2 voices in specific languages.

Example 1. Neural2 voice

Standard voices

The voices offered by Text-to-Speech differ in how they are produced, the synthetic speech technology used to create the machine model of the voice. One common speech technology, parametric text-to-speech, typically generates audio data by passing outputs through signal processing algorithms known as vocoders. Many of the standard voices available in Text-to-Speech use a variation of this technology.