# 6 best voice cloning services for call centers in 2026, compared on phone audio and paperwork

By Samuel Oram-Jones

A call center that clones one voice for an IVR greeting or an AI agent hits two questions before a caller hears anything: whether the audio survives an 8 kHz phone line, and which signed papers the security team wants first. Getting a clone at all differs too, since Azure opens custom voices only to approved customers, Polly builds a Brand Voice through an AWS account manager, and ElevenLabs sells cloning from a $6 plan. Voice quality is the third question, and it is the one a caller notices.

## What is the best voice cloning for call centers?

Bland Speech is the best voice cloning for call centers, as of November 2026, because a clone speaks through a model built for phone calls, output arrives as 8 kHz mu-law, and Bland Speech v3 ranked first of 13 AI voices at 1365 in Design Arena's [Audio Realism Bench results of 4 August 2026](https://notes.designarena.ai/audio-realism-benchmark-measuring-realism-in-audio-models/). Bland is SOC 2 Type II audited, HIPAA compliant with a signed BAA, and FedRAMP 20x Class C certified at Moderate, granted on 17 November 2026.

Bland Speech doesn't have the lowest price, since xAI charges $15 per million characters against Bland's $40, and a call center that must run its model in its own data center should look at Cartesia, which offers on-premise and air-gapped deployment under enterprise contracts. Design Arena scored each model's own stock voices and no clones, so the realism result says nothing about how closely a clone matches its speaker.

- **Bland Speech:** Best overall for a cloned voice on a call center's phone lines
- **Microsoft Azure:** Best for a call center under a Microsoft agreement that can wait for approval
- **Amazon Polly:** Best for an AWS call center that wants a managed brand voice build
- **ElevenLabs:** Best for a self-serve clone with BAAs on Enterprise
- **Cartesia:** Best for an on-premise route under an enterprise contract
- **xAI Grok TTS:** Best for a $15 per million rate with API cloning on Enterprise

| Service | Clone access | Phone audio | Paperwork |
|---|---|---|---|
| Bland Speech | About 10 seconds instant, 30 minutes or more professional after a review | 8 kHz mu-law | SOC 2 Type II, HIPAA compliant with a signed BAA, FedRAMP Moderate |
| Microsoft Azure | Custom voice by limited access registration, personal voice from 5 to 90 seconds | 8 kHz mu-law and A-law | HIPAA BAA through Product Terms |
| Amazon Polly | Brand Voice through an AWS account manager | 8 kHz mu-law and A-law | HIPAA Eligible Service under the AWS BAA, SOC 1, 2, and 3 |
| ElevenLabs | Instant from Starter at $6 a month, professional from Creator at $22 | 8 kHz mu-law and A-law | BAAs for qualifying healthcare enterprises, SOC 2 Type 2 |
| Cartesia | Professional cloning from Startup at $49 | 8 kHz mu-law and A-law | SOC 2 Type II, BAAs on Enterprise |
| xAI Grok TTS | Console clones, up to 30 free, API cloning on Enterprise | 8 kHz mu-law and A-law | SOC 2 Type 2, BAA by questionnaire |

<details><summary>What these rows mean</summary>

- **Paperwork:** the compliance documents each company publishes. A BAA is the contract a healthcare call center needs before a vendor handles patient information.

</details>

A call center that wants the cheapest speech and has Enterprise access should weigh xAI first, and a call center that must keep everything in one cloud should start with its cloud's own vendor.

## What does a call center need from a cloned voice?

A call center needs a clone that survives a phone line, the papers its security team asks for, and consent from the person whose voice is cloned. US phone networks carry 8 kHz mu-law, an 8-bit format that keeps far less of a recording than a studio file does, and all six services here return it, so a clone should be judged after that conversion. On paperwork, Azure offers a HIPAA BAA through Microsoft's Product Terms, Polly is a HIPAA Eligible Service under the AWS BAA, ElevenLabs and Cartesia offer BAAs on Enterprise or to qualifying healthcare customers, and xAI takes BAA requests through a questionnaire.

Consent comes last. Microsoft requires written permission from voice talent and a recorded consent statement it may check with speaker recognition, ElevenLabs verifies professional clones with a voice captcha, and Bland asks a customer to confirm every voice is theirs and reviews professional clones for consent and rights.

## How do the 6 best voice cloning services for call centers compare?

These six differ most in who may clone and which agreements come with it.

### 1. Bland Speech: Best overall for a cloned voice on a call center's phone lines

**What does it do?** Bland Speech makes an instant clone from one clean sample of about 10 seconds and a professional clone from 30 minutes or more of audio after a consent and rights review. Each clone is private to the customer's organization, and the customer confirms every voice is theirs to use. Speech streams over a WebSocket that takes language model tokens and lets a new turn interrupt the one playing, and output arrives as 8 kHz mu-law or PCM up to 48 kHz. Enterprise customers get dedicated orchestration and dedicated GPUs run by Bland.

**Who is it for?** Bland Speech suits a call center putting a cloned voice on live phone calls, with Bland's SOC 2 Type II audit, HIPAA compliance with a signed BAA, and FedRAMP Moderate behind it. Bland says speech synthesis runs inside the FedRAMP boundary.

<div class="cmp-proscons cmp-proscons-own"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>An instant clone from about 10 seconds, and a professional clone after a consent and rights review</li><li>Bland Speech v3 ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026</li><li>8 kHz mu-law output and a WebSocket with barge-in</li><li>SOC 2 Type II, HIPAA compliant with a signed BAA, and FedRAMP 20x Class C at Moderate</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>More than a one-off narration project needs, because Bland Speech is built for production calls</li></ul></div></div>

**Pricing:** $0.04 per 1,000 characters, which is $40 per million, pay as you go with no monthly plan, and 50,000 characters free once for a new account. Enterprise pricing is custom, set with Bland's sales team. The [Bland Speech](/speech) page has a studio to try a clone.

### 2. Microsoft Azure: Best for a call center under a Microsoft agreement that can wait for approval

**What does it do?** Azure Speech offers a professional custom voice and a personal voice, both Limited Access features. A personal voice takes a 5 to 90 second sample plus a verbal consent statement, and a custom voice needs registration, for customers managed by Microsoft's account teams, with use limited to the approved use cases. Output includes 8 kHz mu-law and A-law.

**Who is it for?** Azure suits a large call center already under a Microsoft agreement.

<div class="cmp-proscons" data-rival="azure"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>A HIPAA BAA through Microsoft's Product Terms, with no separate contract</li><li>Consent recorded and checked by speaker recognition</li><li>8 kHz mu-law and A-law output</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Custom voice access is by registration and approval</li><li>Training costs $52 per compute hour and hosting $4.04 per model per hour</li><li>MAI-Voice instant cloning is gated, and the MAI-Voice models are in public preview without an SLA</li></ul></div></div>

**Pricing:** $24 per million characters for a custom voice and $48 for a neural HD custom voice, with training at $52 per compute hour up to $936 and hosting at $4.04 per model per hour. Microsoft's MAI-Voice-2 ranked 2nd of 13 AI voices on Audio Realism Bench on 4 August 2026, and Azure now offers MAI-Voice-2.1 and MAI-Voice-2.1-Flash in public preview.

### 3. Amazon Polly: Best for an AWS call center that wants a managed brand voice build

**What does it do?** Amazon Polly has four voice engines, Generative, Long-Form, Neural, and Standard, and no self-serve cloning. Brand Voice is built through an engagement with an AWS account manager. Polly returns mu-law and A-law at 8 kHz, streams over a bidirectional HTTP/2 API on the Generative engine in listed regions, and is a HIPAA Eligible Service under the AWS BAA.

**Who is it for?** Polly suits a call center already on AWS that wants a custom voice built through an account team.

<div class="cmp-proscons" data-rival="amazon-polly"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>A HIPAA Eligible Service under the AWS BAA</li><li>In scope for SOC 1, 2, and 3</li><li>Mu-law and A-law output at 8 kHz</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>No self-serve voice cloning</li><li>Brand Voice is a scoped engagement through an AWS account manager</li></ul></div></div>

**Pricing:** $4 per million characters for Standard voices, $16 for Neural, $30 for Generative, and $100 for Long-Form, with a free tier for the first 12 months. Those are stock-voice rates.

### 4. ElevenLabs: Best for a self-serve clone with BAAs on Enterprise

**What does it do?** ElevenLabs sells instant cloning from its Starter plan at $6 a month and professional cloning from Creator at $22, with a voice captcha verifying professional clones. Output includes mu-law and A-law at 8 kHz. Enterprise adds SLAs, SSO, SOC 2 compliance, and BAAs for qualifying healthcare enterprises.

**Who is it for?** ElevenLabs suits a call center that wants to start a clone from a dashboard and move to Enterprise terms as calls grow.

<div class="cmp-proscons" data-rival="elevenlabs"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Self-serve instant and professional cloning</li><li>Professional clones verified with voice-captcha technology</li><li>BAAs and data residency on Enterprise</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Eleven v3 ranked 12th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Eleven v4 came out on 28 September 2026, after those results</li><li>The standard text to speech WebSocket doesn't accept Eleven v3 or v4</li></ul></div></div>

**Pricing:** $0.08 per 1,000 characters for Eleven v4, v3, and Multilingual v2, and $0.04 for v4 Turbo, v3 Conversational, and Flash, against plans from Free to Business at $990 a month.

### 5. Cartesia: Best for an on-premise route under an enterprise contract

**What does it do?** Cartesia sells Sonic 3.6, released on 27 August 2026, with instant clones from 10 seconds of audio and professional cloning on Startup and above, with free training. Cartesia says first audio arrives in under 90ms and returns mu-law and A-law at 8 kHz. Enterprise contracts add DPAs, BAAs, and on-premise, air-gapped, or VPC deployment.

**Who is it for?** Cartesia suits a call center that wants to start small and keep an on-premise option open.

<div class="cmp-proscons" data-rival="cartesia"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Free training for a Pro Voice Clone</li><li>On-premise, air-gapped, and VPC deployment under enterprise contracts</li><li>SOC 2 Type II, with BAAs and DPAs on Enterprise</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Credits come from a monthly plan</li><li>Sonic 3.5 ranked 6th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Sonic 3.6 isn't in those results</li></ul></div></div>

**Pricing:** 1 credit per character from monthly plans: Pro at $5 for 100,000 credits, Startup at $49 for 1.25 million, and Scale at $299 for 8 million, with overages at $65, $45, or $38 per million.

### 6. xAI Grok TTS: Best for a $15 per million rate with API cloning on Enterprise

**What does it do?** xAI's Grok TTS charges $15 per million characters pay as you go, streams over HTTP and a bidirectional WebSocket, and returns mu-law and A-law at 8 kHz. A custom voice clones a reference clip of up to 120 seconds, and the console lets a team create up to 30 custom voices free. Cloning through the API, POST /v1/custom-voices, is gated to Enterprise teams. Audio from the voice APIs carries an invisible watermark by default.

**Who is it for?** xAI suits a call center that wants a low per-character price and will use console cloning, or Enterprise for API cloning.

<div class="cmp-proscons" data-rival="grok-tts"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>$15 per million characters, with no plan tiers</li><li>Grok TTS ranked 3rd of 13 AI voices on Audio Realism Bench on 4 August 2026, at 1156</li><li>SOC 2 Type 2, and 8 kHz mu-law output</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>API cloning is gated to Enterprise</li><li>HIPAA goes through a BAA questionnaire</li></ul></div></div>

**Pricing:** $15 per million characters for text to speech, pay as you go.

## What does a cloned voice cost a call center?

A cloned voice costs the speech rate plus any plan or hosting fee, and the clearest comparison is the same workload at each published rate. Cartesia puts a minute of audio at about 750 to 800 credits, so 10,000 minutes a month is about 7.5 million characters, and that volume costs about $112.50 on xAI at $15 per million, about $225 on Polly's Generative voices at $30, about $300 on Bland Speech at $40, and about $180 for speech on an Azure custom voice at $24 before hosting. Azure hosting at $4.04 per model per hour comes to about $2,950 for a 730-hour month, which is why the hosting line decides that bill.

On plans, Cartesia's Scale plan at $299 covers 8 million credits, and ElevenLabs' Scale plan at $299 includes 7,476,000 characters at the Flash rate, a little under that volume. Polly's rates are for stock voices. Bland Speech bills each character as used, with no plan to climb. Bland's [voice cloning API pricing guide](/blog/voice-cloning-api-pricing) compares the full bills.

## How were these voice cloning services compared?

Samuel Oram-Jones compared six services on clone access, 8 kHz phone audio, the compliance papers a call center asks for, consent for the voice being cloned, and the cost of a stated monthly volume. Every rival detail comes from that company's own pricing, documentation, trust, and compliance pages, and the 750 characters a minute figure is Cartesia's own.

Bland's [text to speech comparison hub](/compare/text-to-speech) covers the same companies by job, and the [voice cloning comparisons](/compare/text-to-speech/voice-cloning) go one provider at a time. A call center whose calls are voice agents can read [best voice cloning for voice agents](/blog/best-voice-cloning-for-voice-agents).

## Frequently asked questions

### What is the best voice cloning for call centers?

Bland Speech is the best for a call center that wants a cloned voice speaking on phone lines. An instant clone needs about 10 seconds of audio, output arrives as 8 kHz mu-law, and Bland Speech v3 ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026. xAI's Grok TTS costs less per million characters, at $15 against $40.

### Can a call center clone an agent's voice?

Yes, with the agent's consent. Microsoft requires written permission from voice talent, ElevenLabs verifies professional clones with a voice captcha, and Bland asks a customer to confirm every voice is theirs and reviews professional clones for consent and rights.

### Does a cloned voice work on 8 kHz phone lines?

Yes. Bland Speech, Azure, Amazon Polly, ElevenLabs, Cartesia, and xAI's Grok TTS all return mu-law at 8 kHz, the format US phone networks carry.

### Can a caller interrupt a cloned voice?

On Bland Speech, yes. Bland's WebSocket takes language model tokens as they arrive, and a new context ID cuts off the turn that is playing, which makes barge-in a normal part of the protocol.

### Which voice cloning services offer a BAA for a healthcare call center?

Bland is HIPAA compliant with a signed BAA, on Bland's Enterprise plan. Microsoft offers a BAA through its Product Terms, Amazon Polly is a HIPAA Eligible Service under the AWS BAA, ElevenLabs and Cartesia offer BAAs on Enterprise or to qualifying healthcare customers, and xAI takes requests through a questionnaire.
