# 5 MiniMax speech alternatives in 2026 for cloned voices and phone calls

By Samuel Oram-Jones

A voice cloned on MiniMax's API is deleted if it isn't used within 7 days, and MiniMax bills $1.50 for each cloned voice the first time it speaks. A company that clones its founder for a quarterly campaign, or an agency that keeps a voice per client, can lose a voice between uses, and that rule plus a $100 per million characters rate on the top model gives a team two reasons to price out MiniMax's speech API, as of November 2026.

Five options cover the same job: Bland Speech, ElevenLabs, Cartesia, Google Cloud, and Inworld. Bland Speech comes first because Bland Speech v3 ranked first of 13 AI voices in Design Arena's [Audio Realism Bench results of 4 August 2026](https://notes.designarena.ai/audio-realism-benchmark-measuring-realism-in-audio-models/), and because it bills $40 per million characters with no points to convert. The [text to speech hub](/compare/text-to-speech) and the [alternatives family page](/compare/text-to-speech/alternatives) cover the wider market, and [Bland Speech vs MiniMax](/compare/bland-speech-vs-minimax) and a [MiniMax speech review](/blog/minimax-speech-review) go deeper on MiniMax itself.

## Why do teams look for MiniMax speech alternatives?

Teams look for MiniMax speech alternatives when a clone's lifespan, a points-based package, or the price of the HD model doesn't fit a production voice, even though MiniMax's speech models are capable and cover 40 languages.

MiniMax is a general AI company with speech as one product line beside language, video, image, and music models. Its current speech models are speech-2.8-hd, described as ultra-realistic with sound tags, and speech-2.8-turbo, both part of Speech 2.8, released on 23 January 2026, with speech-2.6 and speech-02 models still sold as legacy. Rapid cloning needs 10 seconds to 5 minutes of audio, voice design builds a voice from a text description for $3, and the API takes 19 inline interjection tags, including laughs, sighs, and breaths, on the 2.8 models only.

The catches are in the billing and the lifecycle. API rates are $100 per million characters for speech-2.8-hd and $60 for speech-2.8-turbo, and audio packages run from $5 a month for 100,000 audio points to $999 for 20 million, with voice slots from 10 on Starter to 500 on Scale and 10% off three-month plans and 20% off yearly ones. The `whisper` and `fluent` emotion values work on 2.6 models, not 2.8, so a team picks between sound tags and those emotions. MiniMax describes speech-2.6-hd as having "Ultra-low latency", and Bland publishes 199ms to first audio.

## What should I check in a MiniMax speech alternative?

A buyer should check how long a cloned voice lasts, what a character costs after plan arithmetic, which features each model supports, and what compliance paperwork the vendor publishes, because those four decide the real cost of a production voice.

Clone lifespan comes first. MiniMax deletes an unused clone after 7 days, so a team should ask every vendor, in writing, what happens to a clone nobody calls for a month, and whether it stays private to the account.

Price arithmetic comes second. ElevenLabs bills dollars per character, Cartesia bills credits, Inworld bills dollars per character with rates that fall by plan, Google prices Gemini voices per token, and Bland Speech bills each character as it's used.

Model features come third, because a vendor's best voice and its fastest voice often differ. Inworld's steering works on TTS-2 only, ElevenLabs' v3 and v4 take audio tags but not SSML break tags, and MiniMax's tags work on speech-2.8 only.

Paperwork comes last. A BAA is the contract HIPAA requires before a vendor handles patient information, and SOC 2 Type II is an independent audit of a vendor's security controls. Bland, ElevenLabs, Cartesia, Google Cloud, and Inworld each state one or both.

## What are the best MiniMax speech alternatives?

The best MiniMax speech alternatives are Bland Speech, ElevenLabs, Cartesia, Google Cloud, and Inworld, with Bland Speech first for phone calls and cloned voices because of its realism result, one standard rate, and phone-ready audio.

| Alternative | Best for | Published price | Audio Realism Bench, 4 August 2026 |
| --- | --- | --- | --- |
| Bland Speech | Cloned voices on phone calls | $40 per million characters, pay as you go | Bland Speech v3: ranked 1st, 1365 |
| ElevenLabs | A large voice library across 90+ languages | $40 to $80 per million characters, by model | Eleven v3: ranked 12th, 938 |
| Cartesia | One model across 44 languages | 1 credit per character, plans from $5 a month | Sonic 3.5: ranked 6th, 1096 |
| Google Cloud | Teams already on Google Cloud | $30 per million characters for Chirp 3: HD, first million free each month | Gemini 2.5 Pro TTS: ranked 5th, 1109 |
| Inworld | A low per-character rate with plans | $15 to $25 per million characters on On-Demand | TTS-1.5 Max: ranked 13th, 763 |

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

- **Published price:** each vendor's own rate with the model or plan named. Inworld's $25 is TTS-2 and $15 is TTS-2 Flash, and paid plans lower both.
- **Audio Realism Bench:** Design Arena's blind listening test, with rank among 13 AI voices and an Elo score. The scored model is named, and Google's row is Gemini TTS.

</details>

### 1. Bland Speech

Bland owns the Bland Speech model and sells it as a studio and an API. POST /v2/tts streams audio over HTTP, a WebSocket at /v2/tts/ws streams text in and audio out, output is mu-law at 8 kHz or PCM from 8 to 48 kHz, and Bland publishes 199ms to first audio. Both endpoints are part of Bland's [text to speech API](/text-to-speech-api).

For a team that clones voices, an instant clone needs one clean sample of about 10 seconds, a professional clone is fine-tuned on 30 minutes or more of audio after Bland's consent and rights review, and every clone is private to the customer's organization, with the customer confirming each voice is theirs to use. Bland is SOC 2 Type II audited, HIPAA compliant with a signed BAA, and FedRAMP 20x Class C certified at Moderate (granted 17 November 2026), and Bland says speech synthesis runs inside that FedRAMP boundary. [Bland Speech vs MiniMax voice cloning](/compare/bland-speech-vs-minimax-voice-cloning) sets the two cloning offers side by side.

<div class="cmp-proscons cmp-proscons-own"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Bland Speech v3 ranked first of 13 AI voices in Audio Realism Bench results of 4 August 2026</li><li>$40 per million characters, below MiniMax's $60 and $100 rates, with no points or packages</li><li>Performance tags such as [laughs] and [say warmly], plus expressiveness and stability controls</li><li>Mu-law at 8 kHz and PCM from 8 to 48 kHz for phone lines</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>More than a hobby project needs, because Bland Speech is built for production calls</li></ul></div></div>

**Pricing:** One standard rate, $0.04 per 1,000 characters ($40 per million), applies in the studio and the API, directed or not, pay as you go with no monthly plan. A new account gets 50,000 free characters once. Enterprise pricing is custom, set with Bland's sales team. The [Bland Speech](/speech) page has a studio for trying a voice.

### 2. ElevenLabs

ElevenLabs launched Eleven v4 and v4 Turbo on 28 September 2026, and the v4 family speaks 90+ languages. The platform suits a team that wants 90+ languages instead of 40 and will manage plan tiers: an instant clone starts on Starter at $6 a month and a professional clone on Creator at $22. ElevenLabs names SOC 2 Type 2, ISO 27001, and HIPAA among its compliance programs, and offers BAAs on Enterprise.

<div class="cmp-proscons" data-rival="elevenlabs"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>90+ languages across the Eleven v4 family</li><li>Audio tags such as [whispering] and [laughing] on Eleven v4</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>Eleven v4 bills $0.08 per 1,000 characters, twice Eleven v4 Turbo's $0.04</li><li>The standard text to speech WebSocket doesn't take Eleven v3 or v4</li></ul></div></div>

**Pricing:** $0.08 per 1,000 characters for Eleven v4 and $0.04 for Eleven v4 Turbo and Flash. Plans run Starter $6 a month, Creator $22, Pro $99, Scale $299, and Business $990, with Enterprise custom.

### 3. Cartesia

Cartesia says Sonic 3.6 delivers first audio in under 90ms across 44 languages. The model suits a team that will manage monthly credit allowances: one credit buys one character, overages are opt-in, and credits roll over up to twice the monthly allotment. An instant clone needs 10 seconds of audio on Pro or above, a professional clone needs 30 minutes on Startup or above, and Cartesia says Sonic can run on-premise or in a customer's own VPC under enterprise contracts.

<div class="cmp-proscons" data-rival="cartesia"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Cartesia says first audio arrives in under 90ms</li><li>SOC 2 Type II, with BAAs and DPAs on Enterprise</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>Credits, rollover limits, and overage settings to manage each month</li><li>Instant cloning needs the Pro plan</li></ul></div></div>

**Pricing:** Free is $0 a month with 20,000 credits, Pro $5 with 100,000, Startup $49 with 1.25 million, and Scale $299 with 8 million, with Enterprise custom. Overages cost $65, $45, and $38 per million credits on Pro, Startup, and Scale.

### 4. Google Cloud

Google Cloud Text-to-Speech has Chirp 3: HD voices and Gemini TTS models, with 380+ voices across 75+ languages and variants. The service suits a team that already buys from Google Cloud, because Text-to-Speech is covered by Google's HIPAA BAA and is in its SOC 2 scope, and it returns mu-law and A-law output. Gemini TTS takes a style prompt and bracket tags such as [sigh] and [short pause].

<div class="cmp-proscons" data-rival="google"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>BAA coverage under Google Cloud's HIPAA agreement</li><li>75+ languages and variants, against 40 at MiniMax</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Chirp 3: Instant Custom Voice is closed to new access requests, and Gemini 3.8 voice replication is in Preview</li><li>Gemini voices bill per token and have no free usage</li></ul></div></div>

**Pricing:** Chirp 3: HD is $30 per million characters after a free first million each month. Gemini 2.5 Pro TTS is $1.00 per million input tokens and $20.00 per million audio tokens, at 25 audio tokens per second of speech.

### 5. Inworld

Inworld sells Realtime TTS-2 and the faster TTS-2 Flash, aimed at consumer apps. Inworld says P90 time to first audio byte is 100ms for TTS-2 and 20ms for Flash, measured server-side and excluding network time. The platform suits a team that wants a lower per-character rate with a monthly plan: instant cloning works from as little as 3 seconds, professional cloning is in beta from at least 10 minutes of audio, and cloned voices bill at the stock-voice rate. Inworld says it is SOC 2 Type II certified and HIPAA and GDPR compliant, with a BAA on Enterprise or as a Growth add-on.

<div class="cmp-proscons" data-rival="inworld"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>TTS-2 Flash at $15 per million characters on On-Demand, falling to $7 on Growth</li><li>Instant clones from 3 seconds, and 50 non-verbal sounds such as [laugh] and [sigh]</li><li>Mu-law and A-law at 8 kHz, and a Twilio ConversationRelay guide</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Steering by plain-language instruction works on TTS-2 only</li><li>Zero data retention starts on the $300 Developer plan and doesn't cover clone samples</li></ul></div></div>

**Pricing:** On-Demand is $0 a month, with TTS-2 at $25 per million characters and TTS-2 Flash at $15. Creator is $25 a month, Builder $100, Developer $300, and Growth $1,500, each returning its fee as credits, with Enterprise custom.

## Which MiniMax speech alternative sounds most human?

Scores in Design Arena's Audio Realism Bench come from blind votes on pairs of recordings, converted to Elo, the rating system chess uses. In its results of 4 August 2026, Bland Speech v3 ranked first of 13 AI voices at 1365, and the human recording scored 1500.

MiniMax's Speech-02 HD ranked 7th at 1093, 272 points behind Bland Speech v3 and three behind Cartesia's Sonic 3.5, which ranked 6th at 1096. Google's Gemini 2.5 Pro TTS ranked 5th at 1109, ElevenLabs' Eleven v3 ranked 12th at 938, and Inworld's TTS-1.5 Max ranked 13th at 763. The model scored for MiniMax was Speech-02 HD and not the current Speech 2.8, and Inworld's TTS-2 and TTS-2 Flash weren't scored either. Cartesia's Sonic 3.6 (27 August 2026) and ElevenLabs' Eleven v4 (28 September 2026) were released after those results.

## Which MiniMax speech alternative should I pick?

For a team cloning voices for phone calls, the clone's lifespan and the bill per character decide it, and both point to Bland Speech: a clone takes one sample of about 10 seconds and stays private to the customer's organization, each character bills at $40 per million with no points to convert, and Bland Speech v3 ranked first of 13 AI voices in those 4 August 2026 results.

<div class="cmp-usewhen"><div class="cmp-usewhen-own"><p><strong>Use Bland Speech when</strong></p><ul><li>A team will clone voices and put them on phone calls with one standard per-character rate</li><li>A team will budget by the character with no points, packages, or voice slots</li></ul></div><div><p><strong>Use another option when</strong></p><ul><li>A team will choose from a large voice library across 90+ languages and manage plan tiers: ElevenLabs</li><li>A team will manage monthly credit allowances and overage settings: Cartesia</li><li>A team already runs on Google Cloud and wants one BAA: Google Cloud</li><li>A team will pick a monthly plan to bring TTS-2 Flash to $7 to $15 per million characters: Inworld</li><li>A team will buy monthly audio points and manage voice slots for 40 languages and sound tags: MiniMax</li></ul></div></div>

Inworld's TTS-2 Flash costs $15 per million characters on On-Demand, a quarter of MiniMax's turbo rate and well below Bland Speech's, so Bland Speech's case rests on realism, phone audio, and one standard rate, not on being the lowest price.

## Frequently asked questions

### Why does MiniMax delete a cloned voice?

MiniMax deletes a cloned voice that isn't used within 7 days, so a seasonal or per-client voice needs a synthesis call inside each week to survive. MiniMax charges $1.50 for a rapid clone the first time the voice is used, not when it is created.

### How much does MiniMax's speech API cost?

MiniMax charges $100 per million characters for speech-2.8-hd and $60 for speech-2.8-turbo, plus $1.50 per cloned voice and $3 per designed voice. Audio packages run $5, $30, $99, $249, and $999 a month, and yearly plans take 20% off.

### What is the best MiniMax speech alternative for phone calls?

Bland Speech is the best MiniMax speech alternative for phone calls, because it returns mu-law at 8 kHz, accepts text from a language model over a WebSocket, and ranked first of 13 AI voices in Design Arena's results of 4 August 2026.

### Does MiniMax's speech API return phone audio?

Yes. MiniMax offers pcmu_raw and pcmu_wav, which are G.711 mu-law at 8 kHz, among its output formats.

### How many languages does MiniMax speech support?

MiniMax's speech-2.6 and speech-2.8 models support 40 languages, and its speech-02 models support 24. ElevenLabs' v4 family speaks 90+, Google Cloud covers 75+ languages and variants, and Cartesia's Sonic 3.6 speaks 44.

### Was Speech 2.8 the model scored in the Audio Realism Bench results?

No. Design Arena's results of 4 August 2026 scored MiniMax Speech-02 HD, released on 2 April 2025, and MiniMax released Speech 2.8 on 23 January 2026.
