Pipecat's list of supported text to speech services has 62 rows, and 31 of them ship inside the Pipecat package, maintained by the Pipecat team, while the other 31 come from the community. For a developer building a phone agent, that split decides which voice survives the next framework update.
Bland Speech is one of the 31 maintained rows, as are the five other services here, each strong at something else: a lower first-audio claim, a lower price, more languages, or running inside a team's own infrastructure.
What is the best text to speech service for Pipecat?
Bland Speech is the best text to speech service for Pipecat on a phone call as of November 2026, on a stated basis: Pipecat's own team maintains its service, the service streams a language model's tokens over a WebSocket and cancels the turn that is playing when a caller interrupts, its voice ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, and it returns mu-law audio for phone lines.
- Bland Speech: Best overall for phone calls where the voice should sound like a person
- Cartesia: Best for a context-based WebSocket and a first-audio claim under 90ms
- ElevenLabs: Best for 90+ languages with a model built for real time
- Deepgram: Best for transcription and speech from one vendor, with self-hosting
- Rime: Best for per-minute billing and a deployment inside a team's own cloud or hardware
- Inworld: Best for the lowest price per character on a fast model
What these rows mean
- How it connects to Pipecat: Pipecat marks a service as maintained when it ships in the Pipecat package and the Pipecat team looks after it.
- Price: Each company's published rate with the plan or model named. Rime bills by the minute and says a minute is about 1,000 characters.
- Phone audio: The 8 kHz formats a phone line carries. Mu-law and A-law are the two standard ways a phone line compresses a voice.
- Audio Realism Bench: Design Arena's blind listening test, where listeners pick which of two voices sounds more like a real person. Scores are Elo ratings, the same kind chess uses, and real human recordings scored 1500.
All six are maintained rows, so the trade is between the voice and two numbers. Cartesia says first audio arrives in under 90ms, Deepgram says as low as 80ms, and Inworld says as low as 50ms, against the 199ms Bland publishes, and Inworld's TTS-2 Flash costs $15 per million characters against Bland's $40. A team that puts a caller's first impression ahead of those numbers has the 4 August 2026 results on its side.
What should a Pipecat team check in a text to speech service?
A Pipecat team should check who maintains the service first. A maintained service ships in the same package as the framework and moves with Pipecat's releases, while a community service depends on one outside maintainer keeping pace.
Streaming comes next, because a voice agent speaks while a language model is still writing. Bland's Pipecat service, BlandTTSService, streams tokens straight to Bland over its WebSocket at /v2/tts/ws and holds the connection open between turns. Its default mode sends tokens as they arrive, and a sentence mode batches text first. Pipecat runs interruptions through Bland's cancel message, and Bland's own protocol also lets a new turn replace the one that is playing.
Voice controls are the last check. Bland's Pipecat settings take a voice, expressiveness and stability values from 0.0 to 1.0, and an auto-formatting switch for phone numbers, all changeable mid-conversation with TTSUpdateSettingsFrame.
Which text to speech services does Pipecat support?
Pipecat maintains 31 services itself, from AWS Polly, Azure, and Google to Bland, Smallest AI, Speechify, and xAI, and 31 more come from the community, among them Murf AI, Respeecher, and Typecast. Pipecat marks the Hume row and an XTTS row as deprecated.
A few of those rows need a second look before a team builds on them. Resemble says it is not selling voice AI to new customers, Hume's homepage and developer docs show no text to speech product, Coqui, the company behind XTTS, shut down in January 2024, and OpenAI has deprecated tts-1, tts-1-hd, and gpt-4o-mini-tts, which are due to shut down on 6 January 2027.
How do the 6 best text to speech services for Pipecat compare?
The six differ in first-audio claims, billing, and where each can run, and the voice itself is where Bland Speech separates from them.
1. Bland Speech: Best overall for phone calls where the voice should sound like a person
What does it do? Bland Speech is the text to speech model inside Bland, a voice AI provider that builds and runs its own speech model. Pipecat ships two Bland classes. BlandTTSService streams over WSS /v2/tts/ws, and BlandHttpTTSService sends complete text to POST /v2/tts, and Pipecat tells voice agents to prefer the streaming class. The install line is uv add "pipecat-ai[bland]", and the Pipecat repository has an example of each. The voice a team passes decides the model, Bland's expressiveness and stability controls are calibrated for its v3 voices, and Bland publishes 199ms to first audio.
Who is it for? Bland Speech suits a team putting a voice on a phone call through Pipecat, where a caller should hear a person and be able to talk over the agent.
Pros
- Bland Speech v3 ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026
- A Pipecat-maintained service with a streaming class and an HTTP class
- One WebSocket held open across turns, with a cancel message for interruptions
- Mu-law at 8 kHz and PCM from 8 to 48 kHz
- $0.04 per 1,000 characters, one standard rate, with no monthly plan
Cons
- More than a hobby project needs, because Bland Speech is built for production calls
Pricing: $0.04 per 1,000 characters, which is $40 per million, with no monthly plan and 50,000 characters free for a new account. Enterprise pricing is custom, set with Bland's sales team. A voice can be auditioned in the Bland Speech studio first.
2. Cartesia: Best for a context-based WebSocket and a first-audio claim under 90ms
What does it do? Cartesia sells Sonic, and Sonic 3.6 was released on 27 August 2026. Its WebSocket works in contexts, and Cartesia says a developer who doesn't have the full transcript up front should use the WebSocket with contexts. Cartesia says first audio arrives in under 90ms, and audio returns as mu-law or A-law at 8 kHz.
Who is it for? Cartesia suits a team that wants one model across 44 languages and will manage a monthly plan of credits.
Pros
- Cartesia says first audio arrives in under 90ms
- 44 languages from Sonic 3.6
- A Pipecat-maintained service
Cons
- Sonic 3.5 ranked 6th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Sonic 3.6 came out after those results
- Credits from a monthly plan
Pricing: 1 credit per character. Pro is $5 a month for 100,000 credits, Startup is $49 for 1.25 million, and Scale is $299 for 8 million, with overages at $65, $45, or $38 per million.
3. ElevenLabs: Best for 90+ languages with a model built for real time
What does it do? ElevenLabs sells text to speech, transcription, and voice agents. Eleven v4 Turbo, released on 28 September 2026, is its real-time model, with a median latency of about 100ms that ElevenLabs states excludes application and network time. Eleven v3 and v4 aren't served by its standard text to speech WebSocket. Audio returns as mu-law or A-law at 8 kHz.
Who is it for? ElevenLabs suits a team that needs many languages and will track which model runs on which endpoint.
Pros
- v4 Turbo at $0.04 per 1,000 characters
- 90+ languages on the Eleven v4 family
- A Pipecat-maintained service
Cons
- Eleven v3 ranked 12th of 13 AI voices on Audio Realism Bench on 4 August 2026
- Eleven v4, its most expressive model, costs $0.08 per 1,000 characters
- A different route than the standard WebSocket for v3 and v4
Pricing: $0.08 per 1,000 characters for Eleven v4, v3, and Multilingual v2, and $0.04 for v4 Turbo and Flash, billed in dollars against plans from Free to Business at $990 a month.
4. Deepgram: Best for transcription and speech from one vendor, with self-hosting
What does it do? Deepgram sells speech recognition, text to speech, and a voice agent API. Flux TTS, its current model, streams over a WebSocket at /v2/speak, and Deepgram says responses start in as low as 80ms. Deepgram publishes a guide for building a real-time voice agent on Pipecat with its transcription and speech together.
Who is it for? Deepgram suits a team already sending call audio to Deepgram for transcription that wants the voice from the same vendor, including one that needs self-hosting.
Pros
- Deepgram says Flux TTS starts in as low as 80ms
- Self-hosted, private cloud, and on-premise deployment for Flux TTS
Cons
- Deepgram's voices weren't in Audio Realism Bench's results of 4 August 2026
- Flux TTS costs $0.045 per 1,000 characters on pay as you go
- SSML is stripped from text instead of read as markup
Pricing: $0.045 per 1,000 characters for Flux TTS, $0.030 for Aura-2, and $0.015 for Aura-1 on pay as you go, with Growth plans from $4,000 a year for lower Aura rates.
5. Rime: Best for per-minute billing and a deployment inside a team's own cloud or hardware
What does it do? Rime sells Coda and Mist v3 for text to speech, aimed at production voice agents and phone lines. Coda's WebSocket takes a language model's tokens and returns word-level timestamps. Audio returns as G.711 mu-law at 8 kHz.
Who is it for? Rime suits a team in healthcare or financial services that wants per-minute billing and a dedicated VPC or its own hardware on an enterprise contract.
Pros
- $50 of free usage every month on the Free plan
- VPC and on-premise deployment on enterprise contracts
- HIPAA compliant with a BAA
Cons
- Rime's voices weren't in Audio Realism Bench's results of 4 August 2026
- Coda serves 10 languages, with one language per voice
- Rime publishes two first-audio figures, 30ms for Coda v2 and a 96ms median for Coda on one request
Pricing: Flex has no monthly fee and charges $0.050 a minute for Coda v2 and $0.030 for Mist v3, while Scale is $100 a month with 20% off every rate. Rime says a minute of speech is about 1,000 characters.
6. Inworld: Best for the lowest price per character on a fast model
What does it do? Inworld sells Realtime TTS-2, launched on 5 May 2026, and Realtime TTS-2 Flash, launched on 9 August 2026, with speech to text and speech to speech beside them. Its WebSocket holds a persistent connection and can detect sentence boundaries as a language model writes, and Pipecat supports it through a built-in InworldTTSService. Inworld says first audio arrives in as little as 50ms for TTS-2 Flash.
Who is it for? Inworld suits a team that wants a low per-character price or is building a consumer voice product such as a companion, a tutor, or a game character.
Pros
- TTS-2 Flash at $15 per million characters on On-Demand
- A built-in Pipecat service
- Mu-law and A-law at 8 kHz
Cons
- TTS-1.5 Max ranked 13th of 13 AI voices on Audio Realism Bench on 4 August 2026, and Realtime TTS-2 wasn't in those results
- Inworld describes its voice AI as built for consumer-facing applications
- Inworld publishes different first-audio figures in different places
Pricing: On-Demand is $0 a month, with Realtime TTS-2 at $25 per million characters and TTS-2 Flash at $15, and up to 70 minutes of speech included.
How were these text to speech services compared?
The six were compared on who maintains the Pipecat service, streaming, interruption, phone audio, published price, and voice realism. Rival details come from each company's own pricing and documentation pages and from Pipecat's list of supported services. Each latency figure is the company's own claim, not a test Bland ran, and voice scores come from Design Arena's published results of 4 August 2026.
Bland's comparison of text to speech APIs covers the same companies by job, and its best lists hold the other rankings. Best text to speech for LiveKit agents ranks the same question for LiveKit, and Best text to speech that handles interruptions compares how each service stops speaking mid-sentence.
Frequently asked questions
What is the best text to speech service for Pipecat?
Bland Speech is the best text to speech service for Pipecat on phone calls. Pipecat's team maintains its service, the service cancels the turn that is playing when a caller interrupts, and its voice ranked first of 13 AI voices in Design Arena's results of 4 August 2026. Cartesia, Deepgram, and Inworld publish lower first-audio claims, and Inworld's TTS-2 Flash costs less per character.
Does Bland Speech work with Pipecat?
Yes. Bland is a text to speech service maintained by the Pipecat team, installed with uv add "pipecat-ai[bland]". It ships a streaming class that uses Bland's WebSocket and an HTTP class for complete text, plus a WebSocket example and an HTTP example in Pipecat's repository.
Which text to speech services does Pipecat support?
Pipecat's list has 62 rows. Thirty-one are maintained by the Pipecat team, among them Bland, Cartesia, ElevenLabs, Deepgram, Rime, Inworld, Azure, Google, and AWS Polly, and the other 31 are community rows.
How does Pipecat handle a caller interrupting the voice?
Pipecat's Bland service sends Bland's cancel message over the open WebSocket, and text under a new context ID replaces the turn that is playing. The client drops any audio that arrived but hasn't played.
How much does Bland Speech cost?
Bland Speech costs $0.04 per 1,000 characters, which is $40 per million, with no monthly plan, and a new account gets 50,000 characters free. Enterprise pricing is custom, set with Bland's sales team.
