# Bland Speech vs Inworld (2026): Which text to speech API is better for phone agents?

Inworld costs less than Bland Speech and says it is faster, so the case for Bland Speech rests on one question: how the voice sounds to a caller. Inworld's Realtime TTS-2 runs $25 per million characters and its Flash model $15, against Bland Speech's $40, and Inworld publishes first-audio figures as low as 50ms against Bland's 199ms.

A team picking a voice for a phone agent has to decide whether those savings are worth more than how the voice lands, and the one independent listening test that scored both companies was published in August 2026.

## Which is better, Bland Speech or Inworld?

For production phone calls where callers should hear a person, Bland Speech is the better choice as of November 2026. Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026 at 1365, and Inworld's TTS-1.5 Max ranked 13th at 763. Inworld's current Realtime TTS-2 wasn't in those results.

Inworld makes more sense for a team that needs a lower price per character, a lower published first-audio figure, or word timings for captions. A team weighing a longer list can start with the [head-to-head comparisons](/compare/text-to-speech/head-to-heads).

<div class="cmp-usewhen"><div class="cmp-usewhen-own"><p><strong>Use Bland Speech when</strong></p><ul><li>A voice agent takes phone calls where a polished, read-aloud voice sounds fake</li><li>A team wants one standard per-character rate and no monthly plan to manage</li><li>A buyer needs speech synthesis to run inside a FedRAMP boundary, with no third-party AI in the call path</li><li>A script needs performed delivery, such as a laugh or a warm read, written inline as tags, with a director in the studio to write the performance</li></ul></div><div><p><strong>Use Inworld when</strong></p><ul><li>A team will pay less per character, at $25 per million on TTS-2 or $15 on TTS-2 Flash, and pick a monthly plan as volume grows</li><li>A product needs word and character timings for captions, or visemes for lip-sync</li><li>A team clones voices from as little as 3 seconds of audio, or serves speech in 200 or more languages and locales</li><li>A phone build runs on Twilio ConversationRelay or Vapi, where Inworld has a guide or an integration</li></ul></div></div>

| Feature | Bland Speech | Inworld |
| --- | --- | --- |
| Audio Realism Bench, 4 August 2026 | Ranked 1st of 13 AI voices, 1365 | TTS-1.5 Max ranked 13th, 763 |
| Price per million characters | $40, one standard rate | $25 on TTS-2, $15 on TTS-2 Flash (On-Demand) |
| Monthly plan | None, pay as you go | On-Demand at $0, paid plans from $25 to $1,500 a month |
| First audio, as each company publishes it | 199ms | As low as 50ms on TTS-2 Flash |
| Phone audio | Mu-law at 8 kHz | Mu-law and A-law at 8 kHz |
| Instant voice clone | About 10 seconds of audio | As little as 3 seconds of audio |
| Delivery controls | Performance tags, expressiveness and stability dials, and timed pauses | Plain-word steering on TTS-2, 50 non-verbal sounds, and SSML breaks |

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

- **Audio Realism Bench, 4 August 2026:** Audio Realism Bench is Design Arena's blind listening test, where native speakers hear two clips of the same line from two models and pick the one that sounds more like a real person. A higher Elo score means listeners picked that voice more often. Inworld's entry was TTS-1.5 Max. Realtime TTS-2 launched on 5 May 2026 and wasn't in those results, and TTS-2 Flash launched on 9 August 2026, after them.
- **Price per million characters:** Bland Speech costs $0.04 per 1,000 characters, so $40 per million, with no subscription. Inworld's On-Demand plan costs $0 a month and bills $25 per million on TTS-2 and $15 on TTS-2 Flash. Paid plans lower the rate, to $12.50 and $7 on Growth at $1,500 a month, and Enterprise goes as low as $5.
- **First audio:** Each figure is the company's own published claim. Inworld says TTS-2 Flash returns first audio in as little as 50ms on a Coval benchmark, and publishes P90 figures of 100ms for TTS-2 and 20ms for TTS-2 Flash, counted on the server and excluding network time. Bland publishes 199ms to first audio on a streamed request.
- **Phone audio:** Both return mu-law at 8 kHz, a compressed telephony format. Inworld also returns A-law.
- **Instant voice clone:** Inworld's clone takes 3 to 30 seconds of audio, and longer samples are trimmed. Bland's instant clone takes one clean sample of about 10 seconds.
- **Delivery controls:** Inworld's plain-word steering works on TTS-2 only, and TTS-2 Flash ignores it. Bland Speech reads tags such as [laughs] and [say warmly] before the words they shape.

</details>

## How do Bland Speech and Inworld compare?

Inworld describes its voice products as realtime voice AI for consumer applications, with solution pages for companions, games, education, and health and wellness, and one for agentic workforce voice agents. Bland built Bland Speech for one setting, phone calls, where in Bland's words polish reads as fake.

### Bland Speech overview

Bland Speech is the text to speech model inside Bland, a voice AI provider that builds and runs its own language, speech, and transcription models for enterprise phone calls. It is sold on its own as an API and a studio on the [Bland Speech](/speech) page.

A team sends text over HTTP to POST /v2/tts, or streams language model tokens into a WebSocket at WSS /v2/tts/ws, where a new context interrupts a turn that is playing. POST /v2/audio/speech follows the OpenAI audio speech format, and a first-party Pipecat service and a LiveKit Agents plugin for Python are available. The [text to speech API](/text-to-speech-api) page has the code.

<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 on Audio Realism Bench on 4 August 2026</li><li>One standard rate, $0.04 per 1,000 characters, with no monthly plan</li><li>Performance tags, a director in the studio, and expressiveness and stability controls</li><li>Mu-law at 8 kHz and PCM from 8 to 48 kHz, ready for phone calls</li><li>Bland runs its own model and speech, and speech synthesis runs inside Bland's FedRAMP boundary</li><li>Clones stay private to the customer's organization, with every voice confirmed as the customer's to use</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>More than a consumer app or a game character needs, because Bland Speech is built for production calls</li></ul></div></div>

### Inworld overview

Inworld sells Realtime TTS-2, which Inworld calls its most powerful model, and Realtime TTS-2 Flash, its fastest and most cost-efficient one. Both stream over HTTP or a persistent WebSocket, where a team can send text as a language model writes it and Inworld detects sentence boundaries on the server. Inworld also serves an OpenAI-format endpoint, and has Pipecat, LiveKit, Vapi, and Twilio ConversationRelay guides or integrations.

Delivery comes from plain-word steering on TTS-2, 50 non-verbal sounds such as [laugh] and [sigh] on both models, SSML break tags, and inline IPA pronunciation. Inworld says Realtime TTS-2 covers 200 or more languages and locales.

<div class="cmp-proscons" data-rival="inworld"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>$25 per million characters on TTS-2 and $15 on TTS-2 Flash, falling to as low as $5 on Enterprise</li><li>First audio as low as 50ms on TTS-2 Flash, by Inworld's own claim</li><li>Instant cloning from as little as 3 seconds of audio, on every plan</li><li>Word and character timings, and visemes for lip-sync</li><li>Mu-law and A-law output for phone lines</li><li>SOC 2 Type II, HIPAA, and GDPR, with a business associate agreement on Enterprise or as a Growth add-on</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>TTS-1.5 Max ranked 13th of 13 AI voices on Audio Realism Bench on 4 August 2026</li><li>Steering works on TTS-2 only, and TTS-2 Flash ignores it</li><li>Professional voice cloning is in beta and runs on TTS-2 only</li><li>Zero data retention starts on the $300 Developer plan and doesn't cover cloning samples</li><li>REST synthesis is capped at 2,000 characters per request</li></ul></div></div>

### How do Bland Speech and Inworld handle phone audio and latency?

Both return audio a phone line can carry, and on published numbers Inworld is faster. Bland publishes 199ms to first audio on a streamed request. Inworld says Realtime TTS-2 Flash returns first audio in as little as 50ms on Coval's benchmark, and TTS-2 in as little as 90ms at P90.

Bland Speech outputs mu-law at 8 kHz and PCM at 8, 16, 24, 44.1, or 48 kHz. Inworld outputs mu-law and A-law at 8 kHz, PCM and WAV from 8 to 48 kHz, MP3, and Opus.

### How do Bland Speech and Inworld handle security?

Inworld says it is SOC 2 Type II certified, HIPAA and GDPR compliant, with a business associate agreement (BAA) on Enterprise or as an add-on on Growth. Inworld offers zero data retention on Developer, Growth, and Enterprise plans, and EU or India data residency on Enterprise.

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's [trust page](/trust-security) says speech synthesis runs inside the FedRAMP boundary and that no third-party AI sits in the call path.

## How much do Bland Speech and Inworld cost?

Inworld costs less per character at every volume on pay as you go. Bland Speech charges $40 per million characters, Inworld's TTS-2 charges $25, and TTS-2 Flash charges $15.

### Bland Speech pricing

Bland Speech costs $0.04 per 1,000 characters, directed or not, in the studio or the API. New accounts get 50,000 characters free, and credits never expire. A 60-second note costs about $0.04, a 10-minute narration about $0.36, and a million characters $40.

Enterprise pricing is custom, set with Bland's sales team.

### Inworld pricing

Inworld's On-Demand plan is $0 a month, pay as you go, with up to 70 minutes of text to speech, 100 custom voices, 2 professional voice clones, and a commercial license. Creator is $25 a month with TTS-2 at $20 per million, Builder is $100 with $17.50, Developer is $300 with $15, and Growth is $1,500 with $12.50. Credits roll over for 3 months. Enterprise is custom, with TTS-2 as low as $5 per million and price matching.

### What does a month of speech cost on each?

At 100,000 characters a month, about two hours of speech, Bland Speech costs $4, and Inworld's On-Demand rates give $2.50 on TTS-2 and $1.50 on TTS-2 Flash. At 1 million characters, about 20 hours, the figures are $40, $25, and $15. At 10 million characters, about 200 hours, they are $400, $250, and $150, before Inworld's higher plans lower the rate further.

A team choosing on price alone chooses Inworld. A team choosing on how a caller hears the voice has the 4 August results to read.

## Which sounds more human, Bland Speech or Inworld?

Bland Speech v3 scored 1365 in the blind tests Design Arena published on 4 August 2026, 602 points above Inworld's TTS-1.5 Max at 763. Bland Speech v3 ranked first of 13 AI voices in those results, TTS-1.5 Max ranked 13th, and a human recording scored 1500.

Inworld's current flagship, Realtime TTS-2, wasn't among the 13 voices, so those scores don't say how it sounds. Listeners hear two clips of the same line, without knowing which model made which, and pick the one that sounds more like a real person. [Design Arena's published results](https://notes.designarena.ai/audio-realism-benchmark-measuring-realism-in-audio-models/) put Bland Speech v3 at an Elo of 1365, Microsoft's MAI-Voice-2 at 1224, and xAI's Grok TTS at 1156.

## Why is Bland Speech the best Inworld alternative?

Bland Speech is the Inworld alternative for a line where the voice on the call matters more than the per-character rate. Bland Speech v3 ranked first among AI voices in an independent blind test on 4 August 2026, and the Inworld model scored beside it ranked last.

Bland runs its own model and speech, with speech synthesis inside its FedRAMP boundary. A team that wants a brand voice can clone one from about 10 seconds of audio, or train a professional clone on 30 minutes or more after Bland's consent and rights review.

Bland's [comparison of text to speech APIs](/compare/text-to-speech) covers more vendors. A team that wants Inworld on its own can read the [Inworld TTS review](/blog/inworld-tts-review) or the [Inworld TTS alternatives](/blog/inworld-tts-alternatives).

## Frequently asked questions

### Is Bland Speech better than Inworld?

For phone calls where callers should hear a person, Bland Speech is better than Inworld. Bland Speech v3 ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026, and Inworld's TTS-1.5 Max ranked 13th. Inworld costs less per character and suits a team that wants the lowest price or word timings.

### Is Bland Speech cheaper than Inworld?

No. Bland Speech costs $40 per million characters, and Inworld's pay as you go rates are $25 on TTS-2 and $15 on TTS-2 Flash. Inworld's paid plans and Enterprise lower its rate further, to as low as $5 per million.

### Which is faster, Bland Speech or Inworld?

On each company's published claim, Inworld. Bland publishes 199ms to first audio on a streamed request, and Inworld says TTS-2 Flash returns first audio in as little as 50ms. Neither number comes from a race run by Bland.

### Which Inworld model did Design Arena score?

Design Arena scored Inworld TTS-1.5 Max, which ranked 13th of 13 AI voices on 4 August 2026. Inworld's current Realtime TTS-2, launched on 5 May 2026, wasn't in those results, and TTS-2 Flash launched on 9 August 2026, after them.

### Can Inworld clone a voice from less audio than Bland Speech?

Inworld's instant clone needs as little as 3 seconds of audio, against about 10 seconds for Bland Speech. Inworld's professional cloning is in beta and needs 10 minutes or more of audio, and Bland Speech's professional clone trains on 30 minutes or more after a consent and rights review.

### Does Inworld work with Twilio, Pipecat, and LiveKit?

Yes. Inworld has a Twilio ConversationRelay guide, a built-in Pipecat service, and a LiveKit Agents plugin. Bland Speech has a first-party Pipecat service and a LiveKit Agents plugin for Python.

### Does Bland Speech work on phone calls?

Yes. Bland built Bland Speech for phone calls, and it returns mu-law at 8 kHz as well as PCM from 8 to 48 kHz, so audio goes onto a phone line without conversion.
