Speechmatics for Voice AI: Speech Capabilities, APIs, Integrations & Implementation
Speechmatics provides enterprise-grade speech APIs: realtime and batch transcription with speaker diarization, low-latency TTS, broad language coverage (55+), SDKs, and first‑party integrations for voice-agent frameworks and no‑code automation.
Peak Demand evaluates the platform in the context of telephony, APIs, business rules, integrations, QA, monitoring, and the operating environment around the agent.
Discuss a Speechmatics DeploymentWhat Is Speechmatics?
Speechmatics is a speech component (STT + TTS) intended as the audio “ear/voice” layer for voice agents and media workflows. It offers realtime websocket and batch transcription (speaker diarization, timestamps, custom vocab), low‑latency TTS, 55+ language support, and documented integrations/SDKs for LiveKit, Pipecat, Vapi and Zapier. It is not a PSTN/telephony provider — telephony features should be implemented with an external telephony/orchestration platform. The Melia multilingual model is documented as production preview and initially batch‑first; check API docs for realtime availability.
Speechmatics Platform Profile
Speechmatics
Speech component • STT/TTS
Where Speechmatics Fits in a Voice AI Technology Stack
Strong component — fits as the STT/TTS layer in voice-agent stacks (via LiveKit/Pipecat/Vapi) and media-transcription pipelines. Good for realtime captioning, multilingual transcription, and deployments that require SaaS, private, or on‑prem options. Poor fit when you need native PSTN trunking, inbound/outbound dialling, IVR dialplans, or an integrated knowledge-retrieval store provided by the speech vendor.
A Typical Speechmatics Production Architecture
The exact architecture depends on the business environment, but Peak Demand evaluates the platform as one layer inside a connected production system.
SpeechmaticsSpeech-to-text and Text-to-speech platform (speech AI stack) intended as the audio 'ear/voice' layer for voice agents and transcription workflows.Speechmatics Capabilities Relevant to Production Voice AI
| Capability | Current position | Scope | Implementation context |
|---|---|---|---|
| Inbound calling | Not applicable | Not applicable / unresolved | Speechmatics provides STT/TTS and realtime APIs / SDKs but does not present native PSTN inbound phone number or call-termination services in the reviewed product docs; telephony must be routed through external telephony/agent platforms. |
| Outbound calling | Not applicable | Not applicable / unresolved | No official docs show native outbound call initiation or PSTN dialing; Speechmatics focuses on audio transcription/synthesis and integrates with agent frameworks for voice pipelines. |
| Telephony / phone routing | Not applicable | Not applicable / unresolved | Routing, IVR dialplans and telephony routing are not documented as native Speechmatics features; such functionality is expected to be handled by external telephony/orchestration platforms. |
| SIP / trunking | Not applicable | Not applicable / unresolved | No first-party SIP trunking or SIP interface documented; realtime connectivity is via websockets/SDKs and integrations (LiveKit/Pipecat) rather than SIP trunks. |
| Webhooks / callbacks | Established | External integration | Speechmatics supports asynchronous webhook delivery patterns via integrations (Zapier Submit Audio uses a webhook URL for async transcript delivery); Batch and job APIs support job lifecycle callbacks via asynchronous flows in integrations. |
| Public APIs | Established | Product-native | Speechmatics publishes REST/WebSocket API references for Realtime (websocket) and Batch transcription and management endpoints (create job, get transcript, usage stats). |
| SDKs / developer libraries | Established | Product-native | Multiple first-party SDKs and voice SDKs are published (Realtime and Batch SDKs for Python/JS/.NET/Rust; TTS SDKs; Pipecat/LiveKit adapters documented). |
| Tool / function calls | Limited / conditional | Platform-family | Speechmatics STT/TTS integrates with agent frameworks (Pipecat, LiveKit, Vapi) that orchestrate LLM calls and tool-like pipelines (examples show OpenAI LLM integration in Pipecat/LiveKit). Speechmatics itself does not claim native LLM 'function' primitives — orchestration is provided by these partner frameworks. |
| Transfers / forwarding / handoff | Not applicable | Not applicable / unresolved | No native call-transfer / human-agent handoff primitives documented; handoff/routing is expected to be implemented in the integrating orchestration platform. |
| Conference / queue primitives | Not applicable | Not applicable / unresolved | Speechmatics documents integrations for multi‑party diarization and realtime agent setups, but it does not provide native telephony queue/conference dialplan services — those are provided by frameworks like LiveKit or external telephony providers. |
| Appointment booking | Not found in reviewed official docs | External integration | Not found in the official documentation reviewed for this profile; this is not a claim that the capability is unsupported. |
| Calendar integration | Not found in reviewed official docs | External integration | Not found in the official documentation reviewed for this profile; this is not a claim that the capability is unsupported. |
| Knowledge bases / retrieval | Not found in reviewed official docs | External integration | Not found in the official documentation reviewed for this profile; this is not a claim that the capability is unsupported. |
| Workflow automation | Established | External integration | Speechmatics supports no-code workflow automation via Zapier (actions for synchronous and async transcripts) and documents integration patterns for pushing transcript outputs to downstream systems. |
| Integrations / connectors | Established | Product-native | First-party integrations and officially documented connectors include LiveKit, Pipecat, Vapi, Zapier and language/SDK adapters; Speechmatics docs provide quickstarts and configuration for these integrations. |
| Call recording | Not found in reviewed official docs | External integration | Not found in the official documentation reviewed for this profile; this is not a claim that the capability is unsupported. |
| Transcription / speech-to-text | Established | Product-native | Core capability: batch and realtime transcription with features including speaker diarization, language identification, custom vocabulary, timestamping and multiple operating points (Standard/Enhanced/Melia). |
| Text-to-speech / voices | Established | Product-native | Speechmatics offers low-latency TTS designed for voice agents, documented TTS SDKs and integration adapters (LiveKit, Pipecat) and pricing per characters; voices and configuration are available in SDK docs. |
| Realtime audio / media streaming | Established | Product-native | Realtime websocket API and realtime SDKs support streaming audio transcription with partial/interim results; integrations with LiveKit and Pipecat enable WebRTC-based realtime agent flows. |
| DTMF / speech gather | Not applicable | Not applicable / unresolved | No documentation of DTMF detection or telephony digit-gathering primitives; DTMF is a telephony feature outside Speechmatics' documented STT/TTS API surface. |
| Call logs / analytics / observability | Established | Product-native | Speechmatics provides usage and job statistics endpoints and documents bolt-on analytics features (topics, sentiment, summaries) as transcription add‑ons; some integrations and SDK examples expose metrics flags for enabling metrics collection. |
| Testing / simulation | Limited / conditional | Platform-family | Speechmatics docs include local quickstarts and test/dev flows within partner frameworks (Pipecat local web client, LiveKit dev modes) and sample/testing guidance (Zapier test steps); native 'simulator' tooling is provided via these integrations rather than as a central Speechmatics test harness. |
| Language support | Established | Product-native | Speechmatics supports 55+ languages and dialects; Melia model adds multilingual code-switching across supported languages. Language detection and hints are documented. |
| Security / compliance | Established | Product-native | Documentation and pricing pages state data encryption in transit and at rest, privacy-first deployment options (SaaS, private, on-prem), and enterprise features such as custom deployments and compliance-ready infrastructure. |
| Pricing / billing model | Established | Product-native | Speechmatics publishes pricing tiers (Free credit starter, Pro, Enterprise), per-hour transcription rates and TTS per-character pricing, concurrency limits per plan, and volume/enterprise discounts in pricing docs. |
Capabilities marked “Not found in reviewed official docs” were not located in the official documentation corpus reviewed for this profile; that status does not mean the capability is unsupported.
How Speechmatics Can Connect to Business Systems
Speechmatics publishes product-native SDKs and first‑party adapters for LiveKit (WebRTC agent flows), Pipecat (pipeline/agent framework), Vapi (voice-agent platform) and Zapier (no-code automation). Realtime connectivity is provided via websocket APIs and WebRTC integrations; orchestration (LLM calls, tool pipelines, queuing, telephony routing) is expected to be implemented in the integrating framework or an external telephony/orchestration provider.
Official, documented integrations include LiveKit (WebRTC agents), Pipecat (pipeline agent framework), Vapi (no-code voice agent platform) and Zapier (no-code automation). (sources: 9,10,3,5,7)
SDKs and adapters exist to embed STT/TTS into agent frameworks (Pipecat/LiveKit) — orchestration (LLM, routing, queuing) is expected to live in those frameworks, not in Speechmatics itself. (sources: 8,9,10,26)
Common Speechmatics Use Cases
Use realtime websocket/SDK paths for low‑latency voice agents with partial/interim results and speaker-aware diarization. Use the Batch API for media transcription jobs and post‑processing workflows. Zapier and webhook patterns are documented for asynchronous transcript delivery to downstream systems. TTS SDKs and LiveKit/Pipecat adapters enable voice‑agent response generation with documented low‑latency voices.
Realtime voice agents (agent orchestration handled by LiveKit/Pipecat/Vapi).
Batch transcription and captioning (media workflows).
Contact center analytics (transcript + bolt-on analytics like topics/sentiment).
No-code automation of transcription via Zapier.
Where Speechmatics May Be Particularly Strong
Key strengths documented in official materials:
Strength 1
Realtime and batch transcription with speaker diarization and turn-detection primitives (developer SDKs & websocket API).
Strength 2
Native TTS with low-latency voices designed for voice agents.
Strength 3
Wide language coverage (55+ languages) and a multilingual code-switching model (Melia).
Strength 4
Multiple integration paths (LiveKit, Pipecat, Vapi, Zapier, SDKs) for rapid agent builds.
Where Speechmatics May Not Be the Best Fit
Important tradeoffs and conditions to validate during selection:
Consideration 1
Not a telephony/PSTN provider — inbound/outbound calling, SIP trunking, DTMF, queuing/transfer primitives are outside the product scope.
Consideration 2
Some advanced models/features (e.g., Melia) are in staged availability (preview / batch-first).
Consideration 3
Orchestration (LLM tool calls, queueing, contact-center routing) relies on partner frameworks rather than being native to the STT/TTS product.
When Peak Demand May Choose Speechmatics
When to choose Speechmatics — and what to confirm during evaluation:
Best-fit pattern 1
Voice AI agents (via LiveKit / Pipecat / Vapi) that need low-latency, multilingual STT + TTS.
Best-fit pattern 2
Realtime captioning, media captioning/transcription pipelines.
Best-fit pattern 3
Enterprises requiring on‑prem or private deployments for privacy.
When another platform may deserve a closer look
Evaluate alternatives when 1
Use cases requiring native telephony trunking, PSTN inbound numbers, call queuing/IVR dialplan features.
Evaluate alternatives when 2
Systems that require an integrated knowledge-base / retrieval store provided by the speech vendor (not present natively).
Security, Data Handling & Compliance Considerations
Security and deployment notes from official docs:
Docs state data encrypted in transit and at rest and offer private/on-prem deployment options for privacy-sensitive workloads. (source: 6)
Realtime API guidance recommends temporary keys for browser-based sessions to avoid exposing long-lived API keys in client contexts. (source: 22)
How Speechmatics Pricing Should Be Evaluated
Pricing model and billing notes documented by Speechmatics:
Free starter credit ($100) and tiered plans (Free / Pro / Enterprise) are published; transcription priced by model and hours, TTS priced per 1k characters in docs. (source: 6)
Pro plan provides higher concurrency and reduced limits; Enterprise offers volume discounts, private/on‑prem deployments and custom pricing. (source: 6)
Testing the Platform Before Production
Testing and staging guidance from the docs:
What Peak Demand Adds Around Speechmatics
Practical implementation checklist and recommended integration path:
Discovery & Platform Fit
Determine whether the platform is actually the right choice for the workflow before building around it.
Conversation & Agent Architecture
Design prompts, flows, variables, tools, validation, escalation and business logic.
Telephony & Realtime Infrastructure
Configure the appropriate phone, SIP, CPaaS or realtime transport layer for the deployment.
Middleware & APIs
Build controlled AWS, Cloudflare, API, webhook or middleware layers where systems require additional validation and orchestration.
Business-System Integration
Connect CRM, scheduling, EMR/EHR, ERP, databases, helpdesk, ordering, field-service or proprietary software where suitable integration surfaces exist.
QA & Managed Operations
Test workflows, monitor production behavior, review failures, measure outcomes and refine the implementation over time.
Speechmatics Questions
What are Speechmatics’ core speech capabilities?
Speechmatics provides realtime and batch transcription with speaker diarization, timestamps, language identification, custom vocabulary, and multiple model operating points (including an enterprise multilingual model called Melia). It also offers native TTS with low‑latency voices and SDKs/adapters for LiveKit, Pipecat and other integrations.
Can I use Speechmatics to build realtime voice agents?
Yes — use the realtime websocket API or the LiveKit/Pipecat adapters to stream audio, receive interim and final transcripts, and synthesize responses with the TTS SDKs. Orchestration (LLM calls, tool invocation, routing/hand‑off) is performed in the integrating framework (LiveKit/Pipecat/Vapi) rather than being a native Speechmatics orchestration layer.
Does Speechmatics provide PSTN inbound/outbound calling or SIP trunking?
Speechmatics is a speech (STT/TTS) platform; telephony primitives such as PSTN inbound numbers, outbound dialing, SIP trunking, DTMF gathering, or IVR dialplans are outside the product’s documented scope and should be implemented with an external telephony provider or via the integrating agent/orchestration platform.
Is the Melia multilingual model available for realtime use?
The Melia multilingual/code‑switching model is documented as a production preview and initially batch‑first. Official docs note staged availability; confirm realtime availability and supported APIs/SDKs in the Speechmatics realtime model documentation before production rollout.
What deployment and privacy options are available?
Official documentation describes SaaS, private and on‑prem deployment options and states data encryption in transit and at rest. Enterprise offerings include custom deployment and compliance‑ready infrastructure; work with Speechmatics sales/engineering to scope private/on‑prem deployments.
How is Speechmatics priced?
Speechmatics publishes tiered pricing (Free starter credit, Pro, Enterprise). Transcription is priced by model and hours; TTS is priced per characters. Documents list concurrency limits per plan and volume/enterprise discounting—verify plan limits relevant to your concurrency and throughput needs.
Does Speechmatics provide built‑in appointment booking, calendar integrations, or call recording?
The official documentation reviewed for this profile does not document native appointment booking, calendar connectors, or built‑in call recording. These workflows are typically implemented by pushing transcripts or events to downstream systems via integrations (Zapier/webhooks) or by pairing Speechmatics with partner platforms.
Explore Retell AI
Retell AI is one of the full-stack Voice AI platforms Peak Demand evaluates for phone-first deployments, custom integrations, telephony, APIs, and managed production workflows.
Explore Retell AIPeak Demand may earn a commission from this link.
Planning a Speechmatics Deployment?
Peak Demand can help evaluate platform fit, design the architecture, connect telephony and business systems, implement controlled tools and integrations, test edge cases, and manage the operational layer after launch.
Discuss a Speechmatics DeploymentOfficial Speechmatics Sources Reviewed
This profile is maintained using official or first-party vendor sources. Current vendor documentation remains the source of truth for an active production decision.
Last researched: 2026-08-20
Next recommended review: 2026-11-18
