Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Murf AI / Speech Gen

Murf Speech Gen2

Proprietary narration-focused speech model with 44.1 kHz native output, pronunciation controls and configurable delivery. Current native access is non-streaming synthesis.

Speech & voice
availableClosed weightsText-to-speech
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
13 of 23 specification fields collected
Text input limitUp to 3,000 characters per TTS requestSee specification notes below
Languages35 listedFull language list in specifications
Input billingSee ratesMurf API
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
Not supportedSource for Streaming (opens in a new tab)

Current native streaming is decommissioned; the non-streaming synthesis model remains available. Historical streaming routes are recorded separately.

Batch processing
Not collected
Fine-tuning
SupportedSource for Fine-tuning (opens in a new tab)

Provider-managed custom voice-model creation, not self-service access to base weights. Instant cloning is a different Falcon 2 capability.

Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Proprietary generative neural speech architecture with a linguistic modeling layerSource for Architecture (opens in a new tab)
Languages
View all 35 entries
  • bg-BG · Bulgarian - Bulgaria
  • bn-IN · Bangla - India
  • de-DE · German - Germany
  • el-GR · Greek - Greece
  • en-AU · English - Australia
  • en-IN · English - India
  • en-SCOTT · English - Scotland
  • en-UK · English - UK
  • en-US · English - US & Canada
  • es-ES · Spanish - Spain
  • es-MX · Spanish - Mexico
  • fr-CA · French - Canada
  • fr-FR · French - France
  • gu-IN · Gujarati - India
  • hi-IN · Hindi - India
  • hr-HR · Croatian - Croatia
  • id-ID · Indonesian - Indonesia
  • it-IT · Italian - Italy
  • ja-JP · Japanese - Japan
  • kn-IN · Kannada - India
  • ko-KR · Korean - Korea
  • ml-IN · Malayalam - India
  • ms-MY · Malay - Malaysia
  • nl-NL · Dutch - Netherlands
  • pa-IN · Punjabi - India
  • pl-PL · Polish - Poland
  • pt-BR · Portuguese - Brazil
  • ro-RO · Romanian - Romania
  • sk-SK · Slovak - Slovakia
  • ta-IN · Tamil - India
  • th-TH · Thai - Thailand
  • tl-PH · Tagalog - Philippines
  • tr-TR · Turkish - Turkey
  • vi-VN · Vietnamese - Vietnam
  • zh-CN · Mandarin - China
Source for Languages (opens in a new tab)

Locales explicitly present in the public model-specific voice table; support varies by voice. Locales and accents are not distinct languages. Truncated locale lists are not expanded by guesswork.

Context caching
Not collected
Published training data
Provider reports 70,000+ hours of ethically sourced speech recordingsSource for Published training data (opens in a new tab)
Audio sample rates
Native 44.1 kHz; API output 8, 24, 44.1 or 48 kHzSource for Audio sample rates (opens in a new tab)

Native model sample rate is documented separately on the Gen2 model page.

Maximum prompt characters
Up to 3,000 characters per TTS requestSource for Maximum prompt characters (opens in a new tab)
Audio formats
MP3, WAV, FLAC, ALAW, ULAW, PCM, OGGSource for Audio formats (opens in a new tab)
Generation controls
Pitch and rate −50 to 50; variation 0–5 (default 1); voice-specific styles; locale; pronunciation dictionary; pause tags; target audioDuration; mono or stereoSource for Generation controls (opens in a new tab)
Voice cloning
Enterprise professional cloning: up to 90 minutes of studio-quality audio, managed turnaround 1–4 weeks; no instant Gen2 cloning is documentedSource for Voice cloning (opens in a new tab)
Voice references
162 unique voice IDs in the reviewed public table; multilingual rows merged by IDSource for Voice references (opens in a new tab)

Access & pricing

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

Murf API

Provider source (opens in a new tab)
API model ID
GEN2
Access
Non-streaming Synthesize Speech API. modelVersion defaults to GEN2; returns an audio URL or Base64 output.
Regions
No endpoint-specific residency commitment collected.
API endpoint URL
https://api.murf.ai/v1/speech/generateSource for API endpoint URL (opens in a new tab)
Streaming
Not supportedSource for Streaming (opens in a new tab)

Current native streaming is decommissioned; the non-streaming synthesis model remains available. Historical streaming routes are recorded separately.

Maximum prompt characters
Up to 3,000 characters per TTS requestSource for Maximum prompt characters (opens in a new tab)
Concurrent requests
5 concurrent non-streaming requests on Free/PAYG; custom enterprise capacitySource for Concurrent requests (opens in a new tab)
Speech timestamps and alignment
Word durations returned; English-only wordDurationsAsOriginalText option preserves unnormalized textSource for Speech timestamps and alignment (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech usageSpeech synthesis1$30.001M characters

Rate conditions

  1. Native pay-as-you-go usage. $10 minimum credit purchase; purchased credits do not expire. Enterprise contracts may differ. Character-based billing is not a text-token rate; free trial credit is a finite allowance.
Pricing source (opens in a new tab)

Data & enterprise terms

Default generated audio URLs are downloadable for 72 hours. encodeAsBase64=true is documented as zero retention of generated audio. This endpoint-specific option does not establish zero retention for account data, clone training inputs or all services.

Read the service policy (opens in a new tab)

Murf API

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

No separate pinned model selector is documented for this service or historical access.

Access
Historical streaming route, now decommissioned. Use Gen2 generate for non-streaming, or Falcon 2 for streaming.
Regions
No endpoint-specific residency commitment collected.
API endpoint URL
https://api.murf.ai/v1/speech/streamSource for API endpoint URL (opens in a new tab)
Streaming
Not supportedSource for Streaming (opens in a new tab)

No longer available for this model; not a statement about current Falcon 2 on the same route.

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

General privacy policy retains personal/account data while the account remains open or needed for services, with stated exceptions after closure. Clone recordings have separate consent, storage and quality-assurance terms. No universal no-training or zero-retention promise is inferred.

Read the service policy (opens in a new tab)

Murf API

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

No separate pinned model selector is documented for this service or historical access.

Access
Historical streaming route, now decommissioned. Use Gen2 generate for non-streaming, or Falcon 2 for streaming.
Regions
No endpoint-specific residency commitment collected.
Streaming
Not supportedSource for Streaming (opens in a new tab)

No longer available for this model; not a statement about current Falcon 2 on the same route.

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

General privacy policy retains personal/account data while the account remains open or needed for services, with stated exceptions after closure. Clone recordings have separate consent, storage and quality-assurance terms. No universal no-training or zero-retention promise is inferred.

Read the service policy (opens in a new tab)

Murf Studio

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

No separate pinned model selector is documented for this service or historical access.

Access
Studio subscription access to Gen2 voice generation. API per-character prices are not Studio plan prices.
Regions
No endpoint-specific residency commitment collected.
API endpoint URL
https://murf.ai/Source for API endpoint URL (opens in a new tab)

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

General privacy policy retains personal/account data while the account remains open or needed for services, with stated exceptions after closure. Clone recordings have separate consent, storage and quality-assurance terms. No universal no-training or zero-retention promise is inferred.

Read the service policy (opens in a new tab)

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.

Murf API

GEN2
Not verified

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How it works
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Minimum reusable input
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Lifetime & refresh
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How to enable it
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Usage fields
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Billing rules
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Limits & data boundaries
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

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

Murf API

Not verified

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How it works
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Minimum reusable input
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Lifetime & refresh
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How to enable it
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Usage fields
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Billing rules
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Limits & data boundaries
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

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

Murf API

Not verified

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How it works
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Minimum reusable input
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Lifetime & refresh
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How to enable it
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Usage fields
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Billing rules
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Limits & data boundaries
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

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

Murf Studio

Not verified

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How it works
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Minimum reusable input
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Lifetime & refresh
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

How to enable it
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Usage fields
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Billing rules
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

Limits & data boundaries
Not collected

No reviewed model-specific discounted prompt-cache contract. WebSocket context IDs and persistent voice profiles are not evidence of prompt caching.

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

Voice choices

162 voices

Choose a voice within Murf Speech Gen2. These are API voice identifiers, not additional model releases. Source links document each voice.

Published voice catalog
Voice / API IDLanguage & accentDelivery
Natalieen-US-natalieen-US (English - US & Canada); fr-FR (French - France); es-ES (Spanish - Spain); it-IT (Italian - Italy); en-UK (English - UK); en-AU (English - Australia); en-IN (English - India); en-SCOTT (English - Scotland); fr-CA (French - Canada); de-DE (German - Germany); es-MX (Spanish - Mexico); pt-BR (Portuguese - Brazil)
English - US & Canada; English - Scotland; French - France; Spanish - Spain; Italian - Italy; Dutch - Netherlands; Tamil - India; Polish - Poland; Bangla - India; Japanese - Japan; Croatian - Croatia; Romanian - Romania; Slovak - Slovakia; Bulgarian - Bulgaria; Czech - Czech Republic; French - Canada; Gujarati - India; Kannada - India; Malayalam - India; Marathi - India; Malay - Malaysia; Punjabi - India; Swedish - Sweden; Telugu - India; Thai - Thailand; Tagalog - Philippines; Vietnamese - Vietnam		Published styles: Angry; Conversational; Furious; Inspirational; Meditative; Narration; Newscast Casual; Newscast Formal; Promo; Sad; Sorrowful; Terrified. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Terrellen-US-terrellen-US (English - US & Canada); es-MX (Spanish - Mexico); es-ES (Spanish - Spain); de-DE (German - Germany)
English - US & Canada; German - Germany; Spanish - Mexico; Spanish - Spain		Published styles: Calm; Conversational; Inspirational; Narration; Promo. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Arianaen-US-arianaen-US (English - US & Canada)
English - US & Canada		Published styles: Conversational; Narration. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Milesen-US-milesen-US (English - US & Canada); es-ES (Spanish - Spain); de-DE (German - Germany); it-IT (Italian - Italy); fr-FR (French - France); en-UK (English - UK); en-AU (English - Australia); en-IN (English - India); en-SCOTT (English - Scotland); fr-CA (French - Canada); es-MX (Spanish - Mexico); pt-BR (Portuguese - Brazil)
English - US & Canada; English - Scotland; French - France; German - Germany; Spanish - Spain; Italian - Italy; Dutch - Netherlands; Polish - Poland; Japanese - Japan; Croatian - Croatia; Slovak - Slovakia; Bulgarian - Bulgaria; Czech - Czech Republic; French - Canada; Gujarati - India; Kannada - India; Malayalam - India; Marathi - India; Malay - Malaysia; Punjabi - India; Swedish - Sweden; Telugu - India; Thai - Thailand; Tagalog - Philippines; Vietnamese - Vietnam		Published styles: Angry; Calm; Conversational; Customer Support Agent; Inspirational; Narration; Newscast; Pirate; Promo; Sad; Sports Commentary; Terrified. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Zionen-US-zionen-US (English - US & Canada)
English - US & Canada		Published styles: Conversational; Narration; Promo. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Amaraen-US-amaraen-US (English - US & Canada); fr-FR (French - France); es-ES (Spanish - Spain); it-IT (Italian - Italy); en-UK (English - UK); en-AU (English - Australia); en-IN (English - India); en-SCOTT (English - Scotland); fr-CA (French - Canada); de-DE (German - Germany); es-MX (Spanish - Mexico); pt-BR (Portuguese - Brazil)
English - US & Canada; English - Scotland; French - France; Spanish - Spain; Italian - Italy; Czech - Czech Republic; Gujarati - India; Kannada - India; Malayalam - India; Marathi - India; Malay - Malaysia; Punjabi - India; Swedish - Sweden; Telugu - India; Thai - Thailand; Tagalog - Philippines; Vietnamese - Vietnam		Published styles: Conversational; Narration; Promo. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Mavericken-US-mavericken-US (English - US & Canada); de-DE (German - Germany); en-UK (English - UK); en-AU (English - Australia); en-IN (English - India); en-SCOTT (English - Scotland); fr-FR (French - France); fr-CA (French - Canada); es-ES (Spanish - Spain); es-MX (Spanish - Mexico); it-IT (Italian - Italy); pt-BR (Portuguese - Brazil)
English - US & Canada; German - Germany; Czech - Czech Republic; Gujarati - India; Kannada - India; Malayalam - India; Marathi - India; Malay - Malaysia; Punjabi - India; Swedish - Sweden; Telugu - India; Thai - Thailand; Vietnamese - Vietnam		Published styles: Narration; Conversational. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.Voice source (opens in a new tab)
Paulen-US-paulen-US (English - US & Canada)
English - US & Canada		Published styles: Audiobook; Conversational; Narration. Some multilingual rows abbreviate additional locales; consult List Voices for the full account-specific inventory.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

Murf reserves proprietary rights to its technology and restricts copying, distribution and reverse engineering. Its speech models are supplied through managed services or privately licensed deployment assets, not an open-weight grant. Service records do not establish independently identified foundation checkpoints.

Classification evidence (opens in a new tab)

Lifecycle dates

Announced streaming deprecation date; non-streaming retained16 Aug 2026Source (opens in a new tab)
Changelog confirms streaming decommissioned5 Oct 2026Source (opens in a new tab)

Sources reviewed