Skip to content

TTS-Seed โ€‹

Updated

This page covers how to call TTS-Seed speech synthesis on MPS. TTS-Seed is plugged into MPS under the engine code tts-s: the APIs, authentication, billing, task queries and storage are all the existing MPS speech-synthesis APIs. You switch to TTS-Seed simply by writing "engine": "tts-s" in the ExtParam extension field, then add emotion, speed and other parameters as needed.

TTS-Seed stands out for having the richest set of emotion / style values: emotion accepts 24 values. Beyond basic emotions such as happy and sad, it includes scenario-specific delivery styles such as storytelling, emotional radio, news, advertising and ASMR whisper.

Voice cloning not supported yet

The TTS-S Engine Integration Guide states explicitly that tts-s does not support cloning yet. Only system voices can be used; get the VoiceId from DescribeVoices.

Choosing between the engines

  • Need scenario-specific delivery styles (news, advertising, storytelling, radio, ASMRโ€ฆ) โ†’ this page
  • Need per-character timestamps for karaoke-style subtitles โ†’ TTS-ElevenLabs
  • Want to describe accent and tone in one natural-language sentence โ†’ TTS-Gemini
  • Need cloning + voice design or independent pitch control โ†’ TTS-MiniMax
  • Need custom emotion tags, or cloning without creating a voice โ†’ TTS-Fishaudio

The full comparison of all seven engines is in MPS TTS Engines Compared.

APIs and Official References โ€‹

All APIs below belong to MPS "Media AI โ€“ Dubbing APIs". Their official references live under the Tencent Cloud International Media AI โ€“ Dubbing APIs directory.

APIs that support tts-s โ€‹

The TTS-S Engine Integration Guide lists the following 4 APIs as switchable to TTS-Seed via ExtParam.engine:

APIPurposeModeOfficial reference
TextToSpeechShort text to speech, returns audio directlySync1041/82311
TextToSpeechAsyncLong text to speech, returns a TaskIdAsync1041/79693
DescribeTextToSpeechAsyncTaskQuery async synthesis result (Status / AudioUrl)Sync1041/79696
DescribeVoicesQuery available voices (including TTS-Seed system voices)Sync1041/79695

Other APIs in the same directory โ€‹

APIPurposeWorks with tts-sOfficial reference
CloneVoiceClone a voice from reference audioโŒ Explicitly unsupported1041/82312
DesignVoiceAsync / DescribeDesignTaskDesign a new voice from a text description / query the resultโณ Not listed in the guide1041/79694 / 1041/79598
UpdateVoice / DeleteVoiceUpdate / delete a cloned or designed voiceโณ Not listed in the guide1041/81214 / 1041/81215
ChangeVoiceVoice conversionโณ Not listed in the guide1041/83750
SyncDubbingEarly all-in-one API (clone or synthesize)โณ Not listed in the guide1041/77775

โŒ = explicitly unsupported in the guide; โณ = not mentioned in the guide โ€” whether it can be switched to TTS-Seed should be confirmed by an actual call.

Common conventions:

  • Version 2019-06-12; endpoint mps.intl.tencentcloudapi.com for the International site (mps.tencentcloudapi.com for the China site).
  • Authentication is TC3-HMAC-SHA256 (SecretId / SecretKey) throughout; see Quick Start for signing.
  • Rate limit is 20 requests/second per API.

Comparison with the other engines โ€‹

MPS currently offers seven external TTS engines. The full side-by-side comparison (API coverage, emotion mechanism, audio parameters, cloning, selection guidance) is on the Blog: MPS TTS Engines Compared โ€” the comparison table is maintained only there, so the engine pages cannot drift apart.

Where tts-s sits: the richest emotion / style enum (24 values, including scenario broadcast styles), and one of only two engines whose sample rate reaches 48000. The trade-off is that cloning is not supported.

Engine Switching: ExtParam.engine โ€‹

The MPS speech-synthesis APIs are engine-agnostic. The ExtParam JSON-string field decides which engine runs the request. tts-s uses two layers:

LayerRoleNotes
engineSelect the engineAlways tts-s on this page
synExt.*MPS-normalized synthesis parametersEmotion, sample rate, duration, speed, volume

Differences from the other engines

The TTS-S Engine Integration Guide lists only the normalized parameters under synExt. It does not list a synExt.engineCustom pass-through layer, nor model / format parameters. If you need these, confirm with an actual call first.

ExtParam is a string in the request, not an object โ€” the JSON must be escaped and passed as a string (see Call Examples).

Main tts-s Parameters โ€‹

ParameterTypeRequiredDescription
engineStringYesSet to tts-s
synExtObjectNoSpeech-synthesis extension parameters
synExt.emotionStringNoEmotion / style of the synthesized speech; see Emotion Values below
synExt.sampleRateIntegerNoSample rate: 8000, 16000, 22050, 24000, 32000, 44100, 48000; default 16000
synExt.durationFloatNoTarget audio duration
synExt.speedFloatNoSpeech rate, range [0.5, 2], default 1.0
synExt.volFloatNoVolume, range (0, 10], default 1.0

Emotion Values โ€‹

synExt.emotion has 24 values, which fall into two groups by purpose:

Basic emotions

ValueMeaningValueMeaning
happyHappysadSad
angryAngrysurprisedSurprised
fearFearfulhateDisgusted
excitedExcitedcoldnessCold / detached
neutralNeutraldepressedDepressed
lovey-doveyCoquettishshyShy
comfortComforting / encouragingtensionRoaring / anxious
tenderTender

Scenario-specific delivery styles

ValueMeaningTypical use
storytellingStorytelling / natural narrationAudiobooks, children's stories
radioEmotional radioLate-night radio, emotional programs
magneticMagneticMale narration, brand promos
advertisingAdvertising / marketingProduct ads, promotional voice-overs
vocal-fryVocal fryDistinctive persona voices
ASMRWhisperSleep aids, ASMR content
newsNews broadcastNews and information
entertainmentEntertainment gossipEntertainment news, short-video commentary
dialectDialectDialect content

Confirm two spellings with an actual call

The guide writes vocal fry as vocal - fry (with spaces around the hyphen) and whisper as ASMR (uppercase). This table records them in conventional enum form as vocal-fry and ASMR; if a call returns a parameter error, retry with the guide's original spelling vocal - fry.

Emotions depend on the voice

Different system voices may support different emotion ranges. After finding a voice with DescribeVoices, try a few emotion values on short text before using them at scale.

Voice (VoiceId) โ€‹

For TTS-Seed, VoiceId takes a TTS-Seed voice name directly, e.g. zh_female_vv_uranus_bigtts. Query available voices with DescribeVoices; see Querying System Voices.

Parameter Example โ€‹

A complete ExtParam object (expanded for readability โ€” escape it to a string when actually sending):

json
{
  "engine": "tts-s",
  "synExt": {
    "emotion": "happy",
    "speed": 0.9
  }
}

Call Examples โ€‹

Speech synthesis TextToSpeech โ€‹

json
{
  "Text": "ๆ‚จๅฅฝ",
  "VoiceId": "zh_female_vv_uranus_bigtts",
  "ExtParam": "{\"engine\":\"tts-s\",\"synExt\":{\"emotion\":\"happy\",\"speed\":0.9}}"
}

Note that the value of ExtParam is an escaped JSON string, not a nested object.

No spaces around the engine value

The guide's original example writes \"engine\":\" tts-s\" (an extra space before tts-s), which is inconsistent with the parameter table and the other examples โ€” a typo. This page corrects it to tts-s. Engine codes are matched as exact strings, so a stray space may prevent switching to TTS-Seed.

Async synthesis and query โ€‹

Submit long text with TextToSpeechAsync; ExtParam is written exactly as for the sync API, and a TaskId is returned. Then poll Status with DescribeTextToSpeechAsyncTask and fetch the audio from AudioUrl once it completes.

Querying System Voices โ€‹

Use DescribeVoices (1041/79695) to list available TTS-Seed voices. ExtParam only needs the engine:

json
{
  "engine": "tts-s"
}

ExtParam is a JSON string for this API as well. It also supports filtering; common input parameters:

ParameterTypeDescription
VoiceTypeStringsystem system voices / clone cloned voices / design designed voices / all everything (default)
GenderStringmale / female / unknown
AgeStringchild / teenager / youth / middle_aged / senior / unknown
Languages.NArray of StringLanguages
Labels.NArray of StringLabels
PageNum / PageSizeIntegerPagination; leave PageSize empty to return all matching entries

Since tts-s does not support cloning, set VoiceType to system to see system voices only.

Limitations and Notes โ€‹

  • Cloning not supported yet: CloneVoice cannot be switched to tts-s; only system voices are available.
  • ExtParam is a string: it must be an escaped JSON string in the request; passing an object is rejected as a parameter error.
  • engine is matched exactly: write tts-s with no leading or trailing spaces.
  • Default sample rate is 16000: for music or high-fidelity scenarios, explicitly raise it to 44100 or 48000.
  • Volume is a half-open interval (0, 10]: 0 is invalid; the minimum must be greater than 0.
  • No model / format / engineCustom listed: the guide gives only the 5 synExt parameters above; whether other parameters take effect should be confirmed by an actual call.
  • โณ-marked APIs are subject to actual calls: the โœ… / โŒ / โณ marks on this page follow what the TTS-S Engine Integration Guide lists; an API not listed is not necessarily unavailable.
DocDescription
MPS TTS Engines ComparedCapabilities, parameters and selection guidance across all seven external engines
AI DubbingOverview of MPS dubbing and built-in engines
TTS-ElevenLabsElevenLabs engine (tts-l1) on the same APIs, with per-character timestamps
TTS-GeminiGoogle Gemini-TTS engine (tts-gg) on the same APIs, style controlled by prompt
TTS-MiniMaxMiniMax engine (minimax) on the same APIs, with cloning and voice design
TTS-FishaudioFish Audio engine (tts-f) on the same APIs, with inline emotion tags and zero-shot cloning
TTS-IndexThe same APIs on IndexTTS (index-tts): a reference audio URL straight in VoiceId, 8-dimension emotion vector
TTS-SeedaudioThe same APIs on Seed Audio (tts-sa): the only engine that generates speech from an image, word-level timestamps
Quick StartAuthentication and signing
Media AI โ€“ Dubbing APIsOfficial API directory

Built with VitePress ยท Deployed on Tencent EdgeOne Pages