Skip to content
Model infoModelsSnapshot: 15 Oct 2026

IBM / Watson Text to Speech

Watson Text to Speech: Enhanced Neural (V3)

Managed enhanced neural (v3) synthesis family with 24 documented voice choices. Voice IDs select speakers and locales, not separately verified foundation-model checkpoints.

Speech & voice
availableClosed weightsText-to-speech
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
13 of 24 specification fields collected
Input content limitPOST / WebSocket: 5 KB text including SSMLSee specification notes below
Languages13 listedFull language list in specifications
Input billingSee ratesIBM Watson Text to Speech
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected
Maximum output
Not collected
Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
AudioSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)

Synthesized audio is emitted in binary chunks over WebSocket after the text request. This is not bidirectional incremental text input.

Batch processing
Not collected
Fine-tuning
Not collected

Provider-assisted custom voice training is documented, but no self-service foundation-model weight fine-tuning contract is verified.

Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Cascaded DNNs for prosody and acoustic features, followed by a neural vocoderSource for Architecture (opens in a new tab)
Languages
View all 13 entries
  • Dutch (Netherlands)
  • English (United Kingdom)
  • English (United States)
  • French (Canadian)
  • French (France)
  • German
  • Italian
  • Japanese
  • Korean
  • Portuguese (Brazilian)
  • Spanish (Castilian)
  • Spanish (Latin American)
  • Spanish (North American)
Source for Languages (opens in a new tab)
Context caching
Not collected
Maximum input content size
POST / WebSocket: 5 KB text including SSMLSource for Maximum input content size (opens in a new tab)

GET: 8 KB total request including URL and headers. Byte limit, not 5,000 characters or tokens.

Output formats
Ogg Opus (default), Ogg Vorbis, MP3 / MPEG, WAV, FLAC, L16 PCM, µ-law, audio/basicSource for Output formats (opens in a new tab)
Audio sample rates
22,050 Hz internal synthesis; format-dependent output resampling from 8 kHz to 192 kHzSource for Audio sample rates (opens in a new tab)

Some formats restrict the allowed rates; default Ogg Opus output is 48 kHz.

Customization options
Language-specific pronunciation dictionaries using IPA/SPR; Premium customers can request a custom voiceSource for Customization options (opens in a new tab)

A pronunciation dictionary is not weight fine-tuning. IBM describes training a custom voice from as little as one hour of supplied audio.

Published price
Standard service pricing starts at USD 0.02 per 1,000 charactersSource for Published price (opens in a new tab)

Published starting price, not a voice-specific fixed quote. Premium/custom voice and Deploy Anywhere terms are quoted separately.

Included free allowance
Lite: 10,000 characters/month for the service instanceSource for Included free allowance (opens in a new tab)

Not a separate allowance per voice or family.

Usage metering
Character billing; whitespace excluded from billable charactersSource for Usage metering (opens in a new tab)

Whitespace and SSML still count toward input byte limits. Exact treatment of every SSML element for billing was not verified here.

Model aliases
View all 14 entries
  • de-DE_BirgitVoice → de-DE_BirgitV3Voice
  • de-DE_DieterVoice → de-DE_DieterV3Voice
  • en-GB_KateVoice → en-GB_KateV3Voice
  • en-US_AllisonVoice → en-US_AllisonV3Voice
  • en-US_LisaVoice → en-US_LisaV3Voice
  • en-US_MichaelVoice → en-US_MichaelV3Voice
  • es-ES_EnriqueVoice → es-ES_EnriqueV3Voice
  • es-ES_LauraVoice → es-ES_LauraV3Voice
  • es-LA_SofiaVoice → es-LA_SofiaV3Voice
  • es-US_SofiaVoice → es-US_SofiaV3Voice
  • fr-FR_ReneeVoice → fr-FR_ReneeV3Voice
  • it-IT_FrancescaVoice → it-IT_FrancescaV3Voice
  • ja-JP_EmiVoice → ja-JP_EmiV3Voice
  • pt-BR_IsabelaVoice → pt-BR_IsabelaV3Voice
Source for Model aliases (opens in a new tab)

Since 2026-03-26, these old V1 selectors route to V3. They are aliases, not newly released V3 models.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

IBM Watson Text to Speech

Provider source (opens in a new tab)
API model ID
Not collected

Select a voice or call the documented feature route; no separate model_id is exposed.

Access
GET/POST /v1/synthesize; choose the voice query parameter from the catalog.
Regions
Use the service-instance URL for the provisioned region. Product-edition and voice-specific availability apply; no universal residency guarantee inferred.
voiceSelectors
View all 24 entries
  • nl-NL_MerelV3Voice
  • en-GB_CharlotteV3Voice
  • en-GB_KateV3Voice
  • en-US_AllisonV3Voice
  • en-US_EmilyV3Voice
  • en-US_HenryV3Voice
  • en-US_KevinV3Voice
  • en-US_LisaV3Voice
  • en-US_MichaelV3Voice
  • en-US_OliviaV3Voice
  • fr-CA_LouiseV3Voice
  • fr-FR_NicolasV3Voice
  • fr-FR_ReneeV3Voice
  • de-DE_BirgitV3Voice
  • de-DE_DieterV3Voice
  • de-DE_ErikaV3Voice
  • it-IT_FrancescaV3Voice
  • ja-JP_EmiV3Voice
  • ko-KR_JinV3Voice
  • pt-BR_IsabelaV3Voice
  • es-ES_EnriqueV3Voice
  • es-ES_LauraV3Voice
  • es-LA_SofiaV3Voice
  • es-US_SofiaV3Voice
Source for voiceSelectors (opens in a new tab)
inputLimit
POST: 5 KB text; GET: 8 KB entire requestSource for inputLimit (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Starting TTS service priceStandard synthesis, starting at1$0.021,000 characters

Rate conditions

  1. Published service-level starting rate, not a separately verified fixed rate for each voice. Whitespace excluded from billing. Lite/Premium/custom-voice and self-hosted plans have separate terms.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

IBM Watson Text to Speech

Provider source (opens in a new tab)
API model ID
Not collected

Select a voice or call the documented feature route; no separate model_id is exposed.

Access
WebSocket /v1/synthesize; choose voice in the connection URL. One text request is synthesized before the service closes the connection.
Regions
Use the service-instance URL for the provisioned region. Product-edition and voice-specific availability apply; no universal residency guarantee inferred.
voiceSelectors
View all 24 entries
  • nl-NL_MerelV3Voice
  • en-GB_CharlotteV3Voice
  • en-GB_KateV3Voice
  • en-US_AllisonV3Voice
  • en-US_EmilyV3Voice
  • en-US_HenryV3Voice
  • en-US_KevinV3Voice
  • en-US_LisaV3Voice
  • en-US_MichaelV3Voice
  • en-US_OliviaV3Voice
  • fr-CA_LouiseV3Voice
  • fr-FR_NicolasV3Voice
  • fr-FR_ReneeV3Voice
  • de-DE_BirgitV3Voice
  • de-DE_DieterV3Voice
  • de-DE_ErikaV3Voice
  • it-IT_FrancescaV3Voice
  • ja-JP_EmiV3Voice
  • ko-KR_JinV3Voice
  • pt-BR_IsabelaV3Voice
  • es-ES_EnriqueV3Voice
  • es-ES_LauraV3Voice
  • es-LA_SofiaV3Voice
  • es-US_SofiaV3Voice
Source for voiceSelectors (opens in a new tab)
inputLimit
5 KB input text including SSMLSource for inputLimit (opens in a new tab)
outputStreaming
Binary audio chunks; append all responses to assemble the output. Optional marks and word timings depend on voice support.Source for outputStreaming (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Starting TTS service priceStandard synthesis, starting at1$0.021,000 characters

Rate conditions

  1. Published service-level starting rate, not a separately verified fixed rate for each voice. Whitespace excluded from billing. Lite/Premium/custom-voice and self-hosted plans have separate terms.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

IBM Watson Text to Speech

Not verified

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

How it works
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Minimum reusable input
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Lifetime & refresh
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

How to enable it
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Usage fields
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Billing rules
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Limits & data boundaries
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

IBM Watson Text to Speech

Not verified

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

How it works
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Minimum reusable input
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Lifetime & refresh
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

How to enable it
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Usage fields
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Billing rules
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

Limits & data boundaries
Not collected

No provider-priced inference cache contract verified. Client-side audio reuse is not provider prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Voice choices

24 voices

Choose a voice within Watson Text to Speech: Enhanced Neural (V3). These are API voice identifiers, not additional model releases. Source links document each voice.

Published voice catalog
Voice / API IDLanguage & accentDelivery
Merelnl-NL_MerelV3VoiceDutch
Netherlands		Female voice; beta availability. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Charlotteen-GB_CharlotteV3VoiceEnglish
United Kingdom		Female voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Kateen-GB_KateV3VoiceEnglish
United Kingdom		Female voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Allisonen-US_AllisonV3VoiceEnglish
United States		Female voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Emilyen-US_EmilyV3VoiceEnglish
United States		Female voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Henryen-US_HenryV3VoiceEnglish
United States		Male voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Kevinen-US_KevinV3VoiceEnglish
United States		Male voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)
Lisaen-US_LisaV3VoiceEnglish
United States		Female voice; generally available. Enhanced Neural (V3) synthesis.Voice source (opens in a new tab)

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

IBM documents this proprietary Watson speech service model or managed voice family. The service exposes recognition/voice selectors, not downloadable trained checkpoints. Closed-weight classification is inferred for the product model; public Granite Speech research releases and referenced open research implementations are separate.

Classification evidence (opens in a new tab)

Sources reviewed