| title | titleSuffix | description | author | manager | ms.service | ms.topic | ms.date | ms.author | ms.custom |
|---|---|---|---|---|---|---|---|---|---|
Text to speech API reference (REST) - Speech service |
Azure AI services |
Learn how to use the REST API to convert text into synthesized speech. |
eric-urban |
nitinme |
azure-ai-speech |
how-to |
9/23/2024 |
eur |
references_regions |
The Speech service allows you to convert text into synthesized speech and get a list of supported voices for a region by using a REST API. In this article, you learn about authorization options, query options, how to structure a request, and how to interpret a response.
Tip
Use cases for the text to speech REST API are limited. Use it only in cases where you can't use the Speech SDK. For example, with the Speech SDK you can subscribe to events for more insights about the text to speech processing and results.
The text to speech REST API supports neural text to speech voices in many locales. Each available endpoint is associated with a region. An API key for the endpoint or region that you plan to use is required. Here are links to more information:
- For a complete list of voices, see Language and voice support for the Speech service.
- For information about regional availability, see Speech service supported regions.
- For Azure Government and Microsoft Azure operated by 21Vianet endpoints, see this article about sovereign clouds.
Important
Costs vary for prebuilt neural voices (called Neural on the pricing page) and custom neural voices (called Custom Neural on the pricing page). For more information, see Speech service pricing.
Before you use the text to speech REST API, understand that you need to complete a token exchange as part of authentication to access the service. For more information, see Authentication.
You can use the tts.speech.microsoft.com/cognitiveservices/voices/list endpoint to get a full list of voices for a specific region or endpoint. Prefix the voices list endpoint with a region to get a list of voices for that region. For example, to get a list of voices for the westus region, use the https://westus.tts.speech.microsoft.com/cognitiveservices/voices/list endpoint. For a list of all supported regions, see the regions documentation.
Note
Voices and styles in preview are only available in three service regions: East US, West Europe, and Southeast Asia.
This table lists required and optional headers for text to speech requests:
| Header | Description | Required or optional |
|---|---|---|
Ocp-Apim-Subscription-Key |
Your Speech resource key. | Either this header or Authorization is required. |
Authorization |
An authorization token preceded by the word Bearer. For more information, see Authentication. |
Either this header or Ocp-Apim-Subscription-Key is required. |
A body isn't required for GET requests to this endpoint.
This request requires only an authorization header:
GET /cognitiveservices/voices/list HTTP/1.1
Host: westus.tts.speech.microsoft.com
Ocp-Apim-Subscription-Key: YOUR_RESOURCE_KEYHere's an example curl command:
curl --location --request GET 'https://YOUR_RESOURCE_REGION.tts.speech.microsoft.com/cognitiveservices/voices/list' \
--header 'Ocp-Apim-Subscription-Key: YOUR_RESOURCE_KEY'
You should receive a response with a JSON body that includes all supported locales, voices, gender, styles, and other details. The WordsPerMinute property for each voice can be used to estimate the length of the output speech. This JSON example shows partial results to illustrate the structure of a response:
[
// Redacted for brevity
{
"Name": "Microsoft Server Speech Text to Speech Voice (en-US, JennyNeural)",
"DisplayName": "Jenny",
"LocalName": "Jenny",
"ShortName": "en-US-JennyNeural",
"Gender": "Female",
"Locale": "en-US",
"LocaleName": "English (United States)",
"StyleList": [
"assistant",
"chat",
"customerservice",
"newscast",
"angry",
"cheerful",
"sad",
"excited",
"friendly",
"terrified",
"shouting",
"unfriendly",
"whispering",
"hopeful"
],
"SampleRateHertz": "24000",
"VoiceType": "Neural",
"Status": "GA",
"ExtendedPropertyMap": {
"IsHighQuality48K": "True"
},
"WordsPerMinute": "152"
},
// Redacted for brevity
{
"Name": "Microsoft Server Speech Text to Speech Voice (en-US, JennyMultilingualNeural)",
"DisplayName": "Jenny Multilingual",
"LocalName": "Jenny Multilingual",
"ShortName": "en-US-JennyMultilingualNeural",
"Gender": "Female",
"Locale": "en-US",
"LocaleName": "English (United States)",
"SecondaryLocaleList": [
"de-DE",
"en-AU",
"en-CA",
"en-GB",
"es-ES",
"es-MX",
"fr-CA",
"fr-FR",
"it-IT",
"ja-JP",
"ko-KR",
"pt-BR",
"zh-CN"
],
"SampleRateHertz": "24000",
"VoiceType": "Neural",
"Status": "GA",
"WordsPerMinute": "190"
},
// Redacted for brevity
{
"Name": "Microsoft Server Speech Text to Speech Voice (ga-IE, OrlaNeural)",
"DisplayName": "Orla",
"LocalName": "Orla",
"ShortName": "ga-IE-OrlaNeural",
"Gender": "Female",
"Locale": "ga-IE",
"LocaleName": "Irish (Ireland)",
"SampleRateHertz": "24000",
"VoiceType": "Neural",
"Status": "GA",
"WordsPerMinute": "139"
},
// Redacted for brevity
{
"Name": "Microsoft Server Speech Text to Speech Voice (zh-CN, YunxiNeural)",
"DisplayName": "Yunxi",
"LocalName": "云希",
"ShortName": "zh-CN-YunxiNeural",
"Gender": "Male",
"Locale": "zh-CN",
"LocaleName": "Chinese (Mandarin, Simplified)",
"StyleList": [
"narration-relaxed",
"embarrassed",
"fearful",
"cheerful",
"disgruntled",
"serious",
"angry",
"sad",
"depressed",
"chat",
"assistant",
"newscast"
],
"SampleRateHertz": "24000",
"VoiceType": "Neural",
"Status": "GA",
"RolePlayList": [
"Narrator",
"YoungAdultMale",
"Boy"
],
"WordsPerMinute": "293"
},
// Redacted for brevity
]The HTTP status code for each response indicates success or common errors.
| HTTP status code | Description | Possible reason |
|---|---|---|
| 200 | OK | The request was successful. |
| 400 | Bad request | A required parameter is missing, empty, or null. Or, the value passed to either a required or optional parameter is invalid. A common reason is a header that's too long. |
| 401 | Unauthorized | The request isn't authorized. Make sure your resource key or token is valid and in the correct region. |
| 429 | Too many requests | You exceeded the quota or rate of requests allowed for your resource. |
| 502 | Bad gateway | There's a network or server-side problem. This status might also indicate invalid headers. |
The cognitiveservices/v1 endpoint allows you to convert text to speech by using Speech Synthesis Markup Language (SSML).
These regions are supported for text to speech through the REST API. Be sure to select the endpoint that matches your Speech resource region.
This table lists required and optional headers for text to speech requests:
| Header | Description | Required or optional |
|---|---|---|
Authorization |
An authorization token preceded by the word Bearer. For more information, see Authentication. |
Required |
Content-Type |
Specifies the content type for the provided text. Accepted value: application/ssml+xml. |
Required |
X-Microsoft-OutputFormat |
Specifies the audio output format. For a complete list of accepted values, see Audio outputs. | Required |
User-Agent |
The application name. The provided value must be fewer than 255 characters. | Required |
If you're using a custom neural voice, the body of a request can be sent as plain text (ASCII or UTF-8). Otherwise, the body of each POST request is sent as SSML. SSML allows you to choose the voice and language of the synthesized speech that the text to speech feature returns. For a complete list of supported voices, see Language and voice support for the Speech service.
This HTTP request uses SSML to specify the voice and language. If the body length is long, and the resulting audio exceeds 10 minutes, it's truncated to 10 minutes. In other words, the audio length can't exceed 10 minutes.
POST /cognitiveservices/v1 HTTP/1.1
X-Microsoft-OutputFormat: riff-24khz-16bit-mono-pcm
Content-Type: application/ssml+xml
Host: westus.tts.speech.microsoft.com
Content-Length: <Length>
Authorization: Bearer [Base64 access_token]
User-Agent: <Your application name>
<speak version='1.0' xml:lang='en-US'><voice xml:lang='en-US' xml:gender='Male'
name='en-US-ChristopherNeural'>
I'm excited to try text to speech!
</voice></speak>* For the Content-Length, you should use your own content length. In most cases, this value is calculated automatically.
The HTTP status code for each response indicates success or common errors:
| HTTP status code | Description | Possible reason |
|---|---|---|
| 200 | OK | The request was successful. The response body is an audio file. |
| 400 | Bad request | A required parameter is missing, empty, or null. Or, the value passed to either a required or optional parameter is invalid. A common reason is a header that's too long. |
| 401 | Unauthorized | The request isn't authorized. Make sure your Speech resource key or token is valid and in the correct region. |
| 415 | Unsupported media type | It's possible that the wrong Content-Type value was provided. Content-Type should be set to application/ssml+xml. |
| 429 | Too many requests | You exceeded the quota or rate of requests allowed for your resource. |
| 502 | Bad gateway | There's a network or server-side problem. This status might also indicate invalid headers. |
| 503 | Service Unavailable | There's a server-side problem for various reasons. |
If the HTTP status is 200 OK, the body of the response contains an audio file in the requested format. This file can be played as it's transferred, saved to a buffer, or saved to a file.
The supported streaming and nonstreaming audio formats are sent in each request as the X-Microsoft-OutputFormat header. Each format incorporates a bit rate and encoding type. The Speech service supports 48-kHz, 24-kHz, 16-kHz, and 8-kHz audio outputs. Each prebuilt neural voice model is available at 24kHz and high-fidelity 48kHz.
amr-wb-16000hz
audio-16khz-16bit-32kbps-mono-opus
audio-16khz-32kbitrate-mono-mp3
audio-16khz-64kbitrate-mono-mp3
audio-16khz-128kbitrate-mono-mp3
audio-24khz-16bit-24kbps-mono-opus
audio-24khz-16bit-48kbps-mono-opus
audio-24khz-48kbitrate-mono-mp3
audio-24khz-96kbitrate-mono-mp3
audio-24khz-160kbitrate-mono-mp3
audio-48khz-96kbitrate-mono-mp3
audio-48khz-192kbitrate-mono-mp3
g722-16khz-64kbps
ogg-16khz-16bit-mono-opus
ogg-24khz-16bit-mono-opus
ogg-48khz-16bit-mono-opus
raw-8khz-8bit-mono-alaw
raw-8khz-8bit-mono-mulaw
raw-8khz-16bit-mono-pcm
raw-16khz-16bit-mono-pcm
raw-16khz-16bit-mono-truesilk
raw-22050hz-16bit-mono-pcm
raw-24khz-16bit-mono-pcm
raw-24khz-16bit-mono-truesilk
raw-44100hz-16bit-mono-pcm
raw-48khz-16bit-mono-pcm
webm-16khz-16bit-mono-opus
webm-24khz-16bit-24kbps-mono-opus
webm-24khz-16bit-mono-opus
riff-8khz-8bit-mono-alaw
riff-8khz-8bit-mono-mulaw
riff-8khz-16bit-mono-pcm
riff-22050hz-16bit-mono-pcm
riff-24khz-16bit-mono-pcm
riff-44100hz-16bit-mono-pcm
riff-48khz-16bit-mono-pcm
Note
If you select 48kHz output format, the high-fidelity voice model with 48kHz will be invoked accordingly. The sample rates other than 24kHz and 48kHz can be obtained through upsampling or downsampling when synthesizing, for example, 44.1kHz is downsampled from 48kHz.
If your selected voice and output format have different bit rates, the audio is resampled as necessary. You can decode the ogg-24khz-16bit-mono-opus format by using the Opus codec.