Speechmatics says its text to speech is "currently in preview and free to use", and that "features and pricing may change as we continue to improve the service". Speechmatics also prices the same service at $0.011 per 1,000 characters, so a developer reading both is looking at a product that has a price and still wears a preview label.
That matters more for a phone agent than for a demo, because a change to an API shows up on a phone line as a failed call.
Which is better, Bland Speech or Speechmatics?
For a live phone line, Bland Speech is the better choice as of November 2026, because it is a launched product and Speechmatics still calls its text to speech a preview. Bland Speech v3 also ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026, and Speechmatics wasn't in those results.
Speechmatics is the cheaper of the two on its list rate, and it makes sense for a team that already buys Speechmatics speech to text and wants speech out and speech in on one balance, in English. A team weighing more vendors than these two can start with Bland's comparison of text to speech APIs.
Use Bland Speech when
- Callers on a phone line should hear a voice that sounds like a person
- A language model's reply should start playing while it is still being written
- An existing OpenAI audio speech integration should move by changing a base URL, a key, and a model name
- A script needs performed delivery, such as a laugh or a warm read, written inline as tags
- A team wants a private instant clone from about 10 seconds of audio, with consent confirmed for every voice
Use Speechmatics when
- A team already runs Speechmatics speech to text and wants text to speech on the same balance
- A team will build on an API where Speechmatics says features and pricing may change during the preview
- A product needs English voices only, in four stock voices named sarah, theo, megan, and jack
- A team will ask Speechmatics sales about running text to speech in its own environment
What these rows mean
- Audio Realism Bench, 4 August 2026: Design Arena's blind listening test, where listeners hear two clips of the same line from two models and pick the one that sounds more like a real person. A higher score means listeners picked that voice more often.
- Status: Speechmatics says text to speech is in preview and free to use, and Speechmatics also prices it, so both statements are on the record.
- Price: Bland Speech costs $0.04 per 1,000 characters in the studio or the API, with no monthly plan. Speechmatics charges $0.011 per 1,000 characters and says one balance covers speech to text, add-ons, and text to speech.
- Streaming: Speechmatics says its API supports streaming audio output but not full bidirectional streaming, and that it plans to add that. Bland Speech takes text over a WebSocket at WSS /v2/tts/ws, where a new context interrupts the turn that is playing.
- First audio: Each figure is the company's own claim, and the two companies describe different things. Speechmatics says the initial audio chunk typically returns in less than 200ms, and also advertises sub-150ms latency.
- Compliance: The Speechmatics line is what Speechmatics states for the company, and Speechmatics adds that it stores input text and generated audio during the text to speech preview.
How do Bland Speech and Speechmatics compare?
The biggest difference is maturity. Bland Speech is a launched product with a studio, an API, and a v3 model that was scored in a public listening test, and Speechmatics text to speech is a preview with four voices and one language that sits beside Speechmatics' speech to text products.
Bland Speech overview
Bland Speech is the text to speech model inside Bland, a voice AI provider that runs its own model and speech rather than wrapping third parties. It is sold on its own as a studio and an API on the Bland Speech page.
A team sends text over HTTP to POST /v2/tts and gets raw audio or a WAV file back, or streams language model output into a WebSocket, where a new context interrupts the turn that is playing. POST /v2/audio/speech follows OpenAI's audio speech contract, and the text to speech API page has the code. Tags in square brackets, such as [laughs], [sighs], or [say warmly], tell the model how to deliver the words after them. Bland publishes 199ms to first audio.
Pros
- Bland Speech v3 ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026
- One standard rate, $0.04 per 1,000 characters, with no monthly plan
- A WebSocket that takes text as a language model writes it
- An OpenAI-compatible endpoint, so an existing integration moves by changing a base URL, a key, and a model name
- Mu-law at 8 kHz and PCM from 8 to 48 kHz, ready for phone calls
- A first-party Pipecat service and a Python plugin for LiveKit Agents
Cons
- More than a hobby project needs, because Bland Speech is built for production calls
Speechmatics overview
Speechmatics says it provides enterprise-grade APIs for speech to text, text to speech, and building voice agents. Its text to speech endpoint sits on a preview host, takes text over HTTP, and returns audio as it is generated. Speechmatics offers four voices and says it supports English, both British and American.
Speechmatics advertises "speaker-aware STT and TTS across 55+ languages", and then says the 55+ languages are for speech to text, so that count belongs to speech to text rather than speech out. Speechmatics has a LiveKit plugin, a Pipecat service, and a Python SDK for text to speech.
Pros
- $0.011 per 1,000 characters, with the first 1 million characters free
- One balance across speech to text, add-ons, and text to speech
- A LiveKit plugin, a Pipecat service, and a Python SDK
- SOC 2 Type II, ISO/IEC 27001:2022, GDPR, and HIPAA for the company, with a DPA and Sub-BAA available
Cons
- Text to speech is "currently in preview", and Speechmatics says features and pricing may change
- Four voices, all English
- Both output formats are 16 kHz
- Text can't stream in as a language model writes it yet, and Speechmatics says it plans to add that
- Speed, pitch, and emphasis controls aren't available yet, per Speechmatics
How do Bland Speech and Speechmatics handle streaming and phone audio?
Both stream audio out, and today Bland Speech is the one that takes text in as it is written. Speechmatics says its API supports streaming audio output but not full bidirectional streaming. A voice agent that pipes a language model's tokens into speech has to collect text before sending it to Speechmatics, while Bland Speech's WebSocket accepts the tokens as they arrive.
On phone formats, Bland Speech returns mu-law at 8 kHz, the 8-bit format built for telephony, as well as PCM at 8, 16, 24, 44.1, or 48 kHz. Speechmatics supports wav_16000 and pcm_16000, which a team would convert before the audio goes onto an 8 kHz phone line.
On speed, each company publishes its own figure. Bland publishes 199ms to first audio, and Speechmatics says the initial chunk typically returns in less than 200ms and advertises sub-150ms latency.
How do Bland Speech and Speechmatics handle voices and security?
Speechmatics offers four stock voices, and Bland Speech adds clones. An instant clone takes one clean sample of about 10 seconds, and a professional clone is fine-tuned on 30 minutes or more of audio and passes Bland's consent and rights review. Every clone is private to the customer's organization, and the customer confirms the 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's trust page says speech synthesis runs inside the FedRAMP boundary.
Speechmatics says it is certified to SOC 2 Type II, ISO/IEC 27001:2022, GDPR, and HIPAA, and offers a DPA and a Sub-BAA. Speechmatics also says that during the text to speech preview it stores input text and generated audio to improve the service.
How much do Bland Speech and Speechmatics cost?
Speechmatics is cheaper on its list rate, and Bland Speech has no preview label on its price. Speechmatics charges $0.011 per 1,000 characters, which is $11 per million, against Bland Speech's $0.04 per 1,000, which is $40 per million.
Bland Speech pricing
Bland Speech costs $0.04 per 1,000 characters, in the studio or the API, with no monthly plan. New accounts get 50,000 characters free, and credits don't 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.
Speechmatics pricing
Speechmatics sells credits, at 1 credit for $1 before discounts, and prices text to speech at $0.011 per 1,000 characters with the first 1 million characters free. New self-serve accounts start with $100 of free credit, and sign-up bonus and credit-pack credits last 12 months. Speechmatics says commercial use and enterprise pricing go through sales, and that pricing may change during the preview.
What does a month of speech cost?
At list rates, 1 million characters a month costs $40 on Bland Speech and $11 on Speechmatics, and 10 million costs $400 and $110. The free first million on Speechmatics lowers its bill further. For a team that mostly needs a low price on English speech, Speechmatics' rate is well under Bland's, and the price of that rate is the preview label, four voices, and audio that arrives at 16 kHz.
Which sounds more human, Bland Speech or Speechmatics?
Speechmatics wasn't in Design Arena's Audio Realism Bench results of 4 August 2026, where Bland Speech v3 ranked first of 13 AI voices at 1365 and real human recordings scored 1500.
- #1BlandBland Speech v31365
- 2ndMicrosoftMAI-Voice-21224
- 3rdxAIGrok TTS1156
- 4thOpenAIGPT Realtime 21119
- 5thGoogleGemini 2.5 Pro TTS1109
- Not rankedSpeechmatics's own voicesNot ranked
Ranks are among 13 AI voices, and Speechmatics's own voices weren't ranked. Real human recordings scored 1500, at the right edge of this scale. Bars start at 600, not zero. Source: Design Arena's published results.
Each listener hears two clips of the same line from two models, without knowing which is which, and picks the one that sounds more like a real person. Speechmatics describes its voices as natural, and that is a description from Speechmatics, not a score. Design Arena publishes the method and the scores in its Audio Realism Bench article.
Why is Bland Speech the best Speechmatics alternative?
On a phone line, Bland Speech brings a launched API with streaming text in, mu-law at 8 kHz, instant and professional clones, and a voice that ranked first of 13 AI voices on Audio Realism Bench on 4 August 2026. The price is higher than Speechmatics' list rate, and what it buys is a product with no preview label.
A team on an OpenAI-style speech client can move by pointing it at Bland's base URL with a Bland key and a Bland model name. Teams comparing other providers can read the list of the best text to speech for 8 kHz phone audio, or Bland's head-to-heads for more vendors.
How was this comparison made?
The criteria are status, voice realism, price, streaming, phone audio, voices, and security. Speechmatics details come from Speechmatics' own text to speech, pricing, and developer documentation pages, and voice scores come from Design Arena's published results of 4 August 2026.
Frequently asked questions
Does Speechmatics sell text to speech?
Yes. Speechmatics sells text to speech at $0.011 per 1,000 characters, and it still calls the service "currently in preview and free to use". Bland Speech is a launched product at $0.04 per 1,000 characters.
Is Speechmatics text to speech cheaper than Bland Speech?
On the list rate, yes. Speechmatics charges $0.011 per 1,000 characters and the first 1 million characters free, and Bland Speech costs $0.04 per 1,000. The difference buys a product that is out of preview, plus mu-law at 8 kHz and text streaming in.
Which languages does Speechmatics text to speech support?
English, both British and American, in four voices. Speechmatics' 55+ languages figure belongs to speech to text.
Does Speechmatics stream text in as a language model writes it?
Not yet. Speechmatics says its API supports streaming audio output but not full bidirectional streaming, and that it plans to add that. Bland Speech's WebSocket takes text in pieces as a language model writes it.
Does Speechmatics output phone audio?
Speechmatics supports two output formats, wav_16000 and pcm_16000. Bland Speech returns mu-law at 8 kHz as well as PCM from 8 to 48 kHz.
Does Speechmatics sound as human as Bland Speech?
Speechmatics wasn't in Design Arena's Audio Realism Bench results of 4 August 2026, so no score from that test compares them. Bland Speech v3 ranked first of 13 AI voices at 1365 in those results.
How fast is Speechmatics text to speech?
Speechmatics says the initial audio chunk typically returns in less than 200ms, and advertises sub-150ms latency. Bland publishes 199ms to first audio.
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.
