Niches · Voice cloning & custom voices

Give a live voice agent a cloned or custom voice

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

0companies, full fit
9companies, fit with a limit
51companies carry the tag
10share of the category's work
4first-hand accounts

What counts as done

The clone's voice id is accepted by the vendor's streaming or websocket TTS and its voice-agent product, every request with that id returns that voice (never a default), the price per character or minute with a clone is stated, and docs say whether the clone works only inside one product.

Measured, for our own tests: first audio from the cloned voice on the streaming endpoint within 500 ms of the first text at the median, and 20 of 20 requests return the cloned speaker. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: make my phone agent use my voice, use the brand voice in the voice agent, stream TTS with my cloned voice, put our cloned voice on the Ultravox agent

Does the job, with a limit 10

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
Cartesia Voice Cloning API + MCP C
Cartesia AI, Inc.
Clones are ordinary voice ids for the WebSocket TTS and Managed Agents, but Pro clones are bound to dated models (sonic-3.5-2026-05-04, sonic-3.6-2026-08-27), and the price per character with a clone is not stated in the pages read. Latency bar unpublished. source
ElevenLabs Voice Cloning and Voice Design API A
Eleven Labs Inc.
Professional clones run on eleven_v4_turbo for conversational agents, but Style and Speed are unavailable on v4, and per-character pricing with a clone was not read here. Latency and 20-of-20 bars unpublished. source
Fish Audio Voice Cloning API B
Hanabi AI Inc.
A model id passes as reference_id to TTS, but acc-03 saw the same request sometimes return a default-sounding voice and acc-04 could not see how to stream with a clone (both closed). Price per byte with a clone not re-read. source
Gradium Voice Cloning and Voice Design B
Gradium SAS
Clones pass as voice_id to the TTS WebSocket stream, shown in the reference examples; sessions last 300 seconds in the FAQ, and the per-character price with a clone was not read. Latency bar unpublished. source
Inworld Instant Voice Cloning B
Theai, Inc. dba Inworld AI
Clones are voice ids for inworld-tts-2, including server streaming, but the guide warns instant clones may not suit children's voices or unusual accents; per-character price with a clone was not re-read. Latency bar unpublished. source
MiniMax Voice Clone and Voice Design C
Nanonoble Pte. Ltd.
Clones work with speech-2.6 and 2.8 hd and turbo, including the T2A WebSocket guide, but the 7-day unused-voice deletion applies, and the flat per-voice fee charged on first use was not re-read. Latency bar unpublished. source
Smallest.ai Instant Voice Clone C
Smallest, Inc.
A clone works with the TTS model in its modelIds, and a clone used on the wrong model returns a 100 to 200 byte WAV rather than an error; lightning-v3.1 clones and its stream routes end on 30 November 2026. Concurrency on self-serve not re-read. source
Soniox Voice Cloning BB
Soniox Inc.
A ready voice id works on the real-time WebSocket (tts-rt-v2) and REST APIs, but voices are prepared only for models that existed at creation and must be recomputed for new ones; 3 concurrent TTS requests by default (listing notes). Latency bar unpublished. source
Speechify API Voice Cloning A
Speechify, Inc.
A clone works in any speech request as soon as it exists, on simba-3.5 (48 languages), simba-3.2 and simba-3.0; simba-3.5 lists at twice the 3.0 rate from 21 November 2026 (listing pricing notes), and streaming latency with a clone is unpublished. source
Ultravox Voice Cloning E
Theai, Inc. dba Inworld AI
The clone is a voice for Ultravox Realtime calls, which is the use, but it works only inside Ultravox and only one clone per account unless support raises it. Latency and 20-of-20 bars unpublished. source

Couldn't check 2

The docs were closed to us, or silent. These carry no mark either way.

ListingWhy
Murf Voice Cloning API C
Murf Inc.
Murf's terms bar automated access to murf.ai, so its API docs were not read this run, as in the 5 November check; no first-hand account names Murf.
Resemble AI Voice Cloning API D
Resemble AI, Inc.
Streaming synthesis pages were not found in the docs index read, and no price per second with a clone was read. source

Doesn't do this job 2

Listings in the category that do something else, with the reason.

ListingWhy
Hume Octave Voice Design and Cloning + MCP C
Hume AI, Inc.
Hume's docs index now lists only Audio Decisions and Expression Measurement, and the voice overview and changelog redirect there; the TTS and EVI APIs, cloning and design included, are no longer documented after the 13 November 2026 end of access. source
VoiceStudio D
VoiceStudio maintainers
Local batch rendering through generate_speech; no streaming TTS endpoint, and acc-10 found a second request was dropped while one rendered. source

What stops agents finishing

  • clone works only inside one product (Ultravox calls)
  • clones usable only with one model or endpoint (Murf Falcon 2 streaming; Gen2 returns 400)
  • one clone per account unless support raises it (Ultravox)
  • 3 concurrent TTS requests by default (Soniox) or one at a time on self-serve (Smallest)

Around it

Before: voice-cloning.clip-to-voice-id. After: text-to-speech.voice-agent-turn, voice-agents.inbound-receptionist.

Where this comes from

Weight 10 of 100 in Voice cloning & custom voices: 4 of 34 accounts: a Fish Audio clone that could not be passed to websocket streaming, a reference_id that sometimes returned a different voice, Cartesia's Pro-clone credit multiplier for an autonomous agent, and a fintech protecting its agents' voices

The same as data: /api/v1/use-cases/voice-cloning.cloned-voice-in-live-agent.json, or the MCP tool get_use_case with id voice-cloning.cloned-voice-in-live-agent. Other jobs in this category: /api/v1/niches?category=voice-cloning.

Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.