# Rime review (2026): What it is and why enterprises choose Bland Speech

By Samuel Oram-Jones

Rime bills text to speech by the minute and Bland Speech bills by the character, so comparing the two takes one conversion, and Rime supplies it: about 1,000 characters a minute. At that rate Rime's Coda v2 costs about $0.05 per 1,000 characters on Flex against Bland Speech's $0.04, and about $0.04 on Rime's Scale plan.

Rime sells conversational voice models to enterprises, and its text to speech competes with Bland Speech, Bland's text to speech model, for regulated phone calls. Facts about Rime are as of November 2026.

## What is Rime and how does it work?

Rime is a voice model company that describes its products as "Voice models made for human conversation," with Prism for speech to speech, Coda and Mist for text to speech in cascade systems, and speech to text. A cascade system chains speech to text, a language model, and text to speech as separate steps. Rime aims its models at production voice agents, IVR (interactive voice response), and telephony in healthcare, financial services, and other regulated industries, and says millions of conversations run through Rime every month.

### Which Rime models are current?

Coda, released in May 2026, is Rime's flagship text to speech model, and Rime says it pairs a language model backbone with a dedicated speech engine trained on full-duplex conversational data. Rime prices the flagship as Coda v2 and mostly calls it Coda elsewhere, with a new on-prem checkpoint released on 28 October 2026. Mist v3, released in March 2026, is the low-latency model.

Prism, a speech to speech model that Rime made available on 28 October 2026, is English only with four voices. Rime retired Arcana from its cloud API on 15 August 2026 and serves those requests with Coda, and Mist v1 and v2 retire on 3 December 2026.

### How does Rime stream audio?

Rime streams text to speech over HTTP, WebSocket, and gRPC. The WebSocket takes language model tokens, and a segment setting of `never`, `bySentence`, or `immediate` controls when synthesis fires. Word-level timestamps are available for Coda over WebSocket and gRPC.

### How fast is Rime, and how fast is Bland Speech?

Rime says Coda v2 reaches first audio at a P50 of 30 ms and a P95 of 60 ms, and Mist v3 at 37 ms and 56 ms at P90, measured at the server without network. Rime's latency benchmark with one request at a time puts Coda at a P50 of 96 ms and a P90 of 98 ms, and at 12 concurrent requests Coda's P50 is 150 ms and its P90 181 ms, while Mist v3 holds at 37 ms and 56 ms. Rime says its cloud API delivers sub-200ms end-to-end latency under typical conditions, and that cloud adds about 25 to 50 ms of network to the model's figure. Bland publishes 199ms to first audio.

On published numbers Rime is faster, each figure as that company publishes it and neither a race Bland ran. Rime's own pages show Coda at 30 ms in one place and 96 ms in another, which is a good reason to measure on a real network.

### What can a Rime voice be told to do?

Rime's custom pauses use angle-bracket markers on Mist v3 only, and Coda handles pauses through punctuation. Coda takes a custom lexicon of X-SAMPA pronunciations (a phonetic notation) in English, German, Spanish, French, Italian, and Portuguese, and Mist v3 takes inline pronunciation on English. A `spell()` function reads letters one by one, and speed is adjustable. Prism, the speech to speech model, reasons over a caller's tone as well as the words.

Bland Speech v3 takes performance tags written in the text, such as [laughs], [sighs], and [say warmly], plus controls for expressiveness and stability from 0.0 to 1.0 and timed pauses.

### Which languages and voices does Rime have?

Coda's cloud API serves 10 languages, English, Arabic, Dutch, French, German, Hindi, Italian, Japanese, Portuguese, and Spanish, and each voice serves one language. Coda has 313 voices, 185 of them English and 1 Dutch. Mist v3 has 78 voices in English, French, German, and Spanish.

### What phone audio does Rime output?

Rime outputs G.711 mu-law (`audio/PCMU`) and says to set `samplingRate: 8000` to match a telephony stream, with Opus, MP3, WAV, and headerless 16-bit PCM also offered and a default sampling rate of 24000. Bland Speech returns 8-bit mu-law at 8 kHz for telephony, and PCM at 8000, 16000, 24000, 44100, or 48000 Hz.

## How much does Rime cost?

Rime has four plans, Free at $0 with $50 of usage a month, Flex with no monthly fee billed by the minute, Scale at $100 a month with 20% off every rate, and a custom Enterprise plan billed annually, while Bland Speech has no plan and charges $0.04 per 1,000 characters ($40 per million). Rime says a minute of text to speech is about 1,000 characters.

On Flex, Mist v3 costs $0.030 a minute, Coda v2 $0.050, and Prism $0.080, and speech to text is $0.002 a minute. On Scale, the rates drop to $0.024, $0.040, and $0.064, and the $100 of credit rolls over up to $200. Enterprise brings custom volume pricing, a deployment in the cloud, a VPC, or on-prem, an SLA with a forward deployed engineer and a linguist, a BAA, a SOC 2 Type II report, and custom voices.

| | Bland Speech v3 | Rime Coda v2 |
| --- | --- | --- |
| Price per million characters | $40 | About $50 on Flex, about $40 on Scale |
| Billing unit | Per character | Per minute of audio, at about 1,000 characters a minute |
| Free start | 50,000 characters, once | $50 of usage every month |
| First audio, as each company publishes it | 199ms | 30 ms at P50 on the server, or 96 ms in a one-request benchmark |
| Voice cloning | Instant from about 10 seconds, professional from 30 minutes | Enterprise clone ordered from Rime, from 30 to 60 minutes of recordings |
| Phone encoding | Mu-law at 8 kHz | G.711 mu-law at 8000 Hz sampling |

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

- **Price per million characters:** Bland Speech is $0.04 per 1,000 characters, which is $40 per million, in the studio or the API. Rime's Coda v2 is $0.050 a minute on Flex and $0.040 on Scale, which becomes about $50 and about $40 per million characters at Rime's own conversion of about 1,000 characters a minute.
- **Billing unit:** Rime bills the length of the audio, so the character equivalent moves with speaking pace. Bland Speech bills the characters sent.
- **Free start:** Bland gives a new account 50,000 free characters once. Rime's Free plan renews $50 of usage every month.
- **First audio:** Rime says 30 ms at P50 measured at the server, and Rime's latency benchmark gives 96 ms for one request at a time. Bland publishes 199ms to first audio on a streamed request.
- **Voice cloning:** Bland Speech makes an instant clone from one clean sample of about 10 seconds and a professional clone from 30 minutes or more. Rime says to plan for 30 to 60 minutes of clean recordings for a basic enterprise clone.

</details>

On Flex, Rime's Coda v2 comes to about $50 per million characters against $40 for Bland Speech, and on Scale the two match at about $40. Rime's Mist v3 is cheaper, at about $30 per million on Flex and about $24 on Scale, and a team that picks it trades the flagship model for the saving.

### How is Bland Speech priced?

Bland Speech costs $0.04 per 1,000 characters, one standard rate in the studio and the API, pay as you go with no monthly plan, and a new account gets 50,000 characters free once, which is nearly an hour of speech. Credits do not expire. Enterprise pricing is custom, set with Bland's sales team.

## How easy is Rime to use?

Rime integrates with LiveKit, Pipecat, Daily, and Vapi, says framework integrations come on every plan, and offers SDKs for Python, Node.js, Go, and Rust (the Prism SDK), plus a CLI and a hosted MCP server. Rime's LiveKit plugin covers both Prism and text to speech, and deployment guides exist for Baseten and Cerebrium. Prism uses the OpenAI Realtime beta event names.

Bland Speech has a first-party Pipecat service, maintained by the Pipecat team, a LiveKit Agents plugin for Python, and an OpenAI-compatible speech endpoint, all described at the [text to speech API](/text-to-speech-api).

### How does cloning compare?

Rime offers enterprise voice cloning, ordered from Rime and trained on Coda or Mist, from 30 to 60 minutes of clean recordings for a basic clone and two to five hours or more for the strongest results. Bland Speech makes an instant clone from one clean sample of about 10 seconds, ready in about a minute, and a professional clone fine-tuned on 30 minutes or more of audio after a consent and rights review. Every clone is private to the customer's organization, and the customer confirms the voice is theirs to use.

## Is Rime secure enough for enterprise calls?

Rime is built for regulated calls, and Rime says it is SOC 2 Type 2 compliant as of May 2025 and HIPAA compliant as of February 2024, with its most recent audit completed in March 2026. Rime says it will sign an MSA, a BAA, or DPAs, and that it never uses customer text or audio to train its models. By default Rime keeps no customer data beyond a character count.

Rime says enterprise customers can run Rime in a dedicated VPC or fully on-premises, and Rime's on-prem engine runs on an NVIDIA host with Docker Compose.

### How does Bland handle security?

Bland is SOC 2 Type II audited, HIPAA compliant with a signed BAA (a business associate agreement, the contract HIPAA requires with a vendor that handles health data), and FedRAMP 20x Class C certified at Moderate, granted on 17 November 2026, and Bland states that speech synthesis runs inside the FedRAMP boundary. The BAA sits on Enterprise, where customers also get dedicated orchestration and dedicated GPUs run by Bland. Enterprise pricing is custom, set with Bland's sales team.

## Is Rime worth it for enterprises?

Bland Speech is the stronger choice for an enterprise phone line: its v3 model ranked first of 13 AI voices in Design Arena's results of 4 August 2026, which didn't include Rime, and a team clones a voice from 10 seconds of audio instead of ordering one built from 30 to 60 minutes of recordings.

### Rime pros and cons

<div class="cmp-proscons" data-rival="rime"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Rime says Coda v2 reaches first audio at a P50 of 30 ms at the server</li><li>Mist v3 costs $0.030 a minute on Flex, about $30 per million characters</li><li>SOC 2 Type 2 and HIPAA compliance, a BAA, and a dedicated VPC or on-prem option on Enterprise</li><li>HTTP, WebSocket, and gRPC streaming, with word-level timestamps for Coda</li><li>313 Coda voices, G.711 mu-law output, and a free $50 of usage every month</li></ul></div><div class="cmp-proscons-cons"><p><strong>Cons</strong></p><ul><li>Custom pauses work on Mist v3 only, and Coda reads pauses from punctuation</li><li>Voice cloning is an Enterprise order that needs 30 to 60 minutes of recordings</li><li>Billing is per minute of audio, so the cost per character moves with speaking pace</li></ul></div></div>

### Why enterprises choose Bland Speech over Rime

Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, at 1365. Rime wasn't in those results, so no score there places a Rime voice beside Bland's. In [Design Arena's published results](https://notes.designarena.ai/audio-realism-benchmark-measuring-realism-in-audio-models/), vetted native speakers heard two clips of the same transcript, without knowing which model made which, and chose the one that sounded more like a real person.

Bland owns its speech model, language model, and transcription, so one company answers for a whole call. Bland bills per character with no plan to track, and Bland's WebSocket takes text as a language model writes it, with a new context interrupting the turn that is playing. A team can read more in [Bland Speech vs Rime](/compare/bland-speech-vs-rime), [Rime alternatives](/blog/rime-alternatives), and [Bland Speech vs Deepgram vs Rime](/blog/bland-speech-vs-deepgram-vs-rime).

<div class="cmp-proscons cmp-proscons-own"><div class="cmp-proscons-pros"><p><strong>Pros</strong></p><ul><li>Ranked first of 13 AI voices in Design Arena's results of 4 August 2026, at 1365</li><li>One standard rate of $0.04 per 1,000 characters, pay as you go, with 50,000 characters free for a new account</li><li>Performance tags such as [laughs] and [say warmly] written in the text</li><li>Instant cloning from about 10 seconds of audio</li><li>FedRAMP 20x Class C certified at Moderate, with speech synthesis inside the boundary</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>

### Verdict: Rime or Bland Speech for phone calls?

Bland Speech is the one to buy for phone calls, because its bill is a per-character figure a team never converts from minutes, and it comes in at or under Coda v2 on Rime's Flex and Scale plans. Rime suits an IVR team that wants Mist v3's lower per-minute rate. A team can start from the [text to speech hub](/compare/text-to-speech) or the [review family](/compare/text-to-speech/reviews), and try [Bland Speech](/speech) on a real script.

## Frequently asked questions

### What is Rime?

Rime is a voice model company that sells Coda and Mist for text to speech, Prism for speech to speech, and speech to text, aimed at production voice agents, IVR, and telephony in regulated industries.

### How much does Rime cost?

Rime's Coda v2 costs $0.050 a minute on Flex and $0.040 on Scale, and Mist v3 costs $0.030 and $0.024, with a Free plan of $50 of usage a month. At Rime's conversion of about 1,000 characters a minute, Coda v2 is about $50 and $40 per million characters, against $40 for Bland Speech.

### Is Bland Speech cheaper than Rime?

Bland Speech costs less than Coda v2 on Flex, matches Coda v2 on Scale, and costs more than Mist v3, which is about $30 per million characters on Flex and about $24 on Scale. Bland Speech holds the 4 August 2026 realism result and one standard per-character rate.

### Does Rime support voice cloning?

Yes, Rime offers enterprise voice cloning on Coda or Mist, ordered from Rime, from 30 to 60 minutes of clean recordings for a basic clone. Bland Speech makes an instant clone from about 10 seconds and trains a professional clone on 30 minutes or more.

### How fast is Rime?

Rime says Coda v2 reaches first audio at a P50 of 30 ms at the server, and Rime's one-request benchmark puts Coda at 96 ms. Bland publishes 199ms to first audio, and each figure is as that company publishes it.

### What is the best Rime alternative for phone calls?

Bland Speech is the option to put beside Rime for phone calls, since Bland Speech v3 ranked first of 13 AI voices in Design Arena's results of 4 August 2026 and returns 8 kHz mu-law audio. More options are in [Rime alternatives](/blog/rime-alternatives).
