Speechmatics is a strong pick when you need accurate live captions, speaker labels, or an on-prem deployment path. The gap shows up when the product also needs a second language on the same screen: realtime translation is a paid add-on on Speechmatics, and the language pairs are English-anchored. The LecSync Realtime Speech API returns captions, translation and speaker labels on one WebSocket for $0.006 per audio minute.
This comparison uses public sources only. Speechmatics figures come from the pricing page, the real-time product page, and the translation docs, read on October 11, 2026. LecSync figures come from the developers page and the pricing docs.
The short answer
If you need live captions, a translation track and speaker labels together, LecSync returns all three over one connection at $0.36 per hour (60 × $0.006). Speechmatics Realtime Standard lists at $0.45 per hour and Enhanced at $0.80 per hour; speaker diarization is included. Realtime translation is an add-on at $0.65 per hour, so Standard plus translation is $1.10 per hour and Enhanced plus translation is $1.45 per hour.
Speechmatics is the better fit when you need on-prem or on-device deployment, published compliance certifications, up to five English-paired translation targets on one job, a custom dictionary, a medical model, or a voice-agent model with turn detection.
What you wire up on each side
With LecSync, your server makes one POST to /v1/realtime/connect and gets back a stream URL plus a one-time token. The client opens the WebSocket and pushes 16 kHz mono PCM. Transcript segments, their translations, speaker labels and optional live minutes all come back on that same socket, already paired.
With Speechmatics, you open a realtime WebSocket to their ASR endpoint, send audio, and receive partials and finals. Set diarization to speaker for speaker labels. To add translation, attach a translation_config with up to five target languages; translation messages arrive after transcription messages and do not block captions. Summaries, chapters, topics and sentiment are separate add-ons and, on the public pricing page, are listed for pre-recorded jobs rather than live streams. We covered the one-connection pattern in more depth in built-in translation vs a DIY STT + MT pipeline.
Side-by-side facts (checked October 11, 2026)
| LecSync Realtime Speech API | Speechmatics Realtime | |
|---|---|---|
| Connection | One WebSocket after POST /v1/realtime/connect | WebSocket to Speechmatics realtime ASR |
| Live captions | Included | Standard $0.45/hr, Enhanced $0.80/hr, Linden 1 $0.30/hr (PAYG list) |
| Live translation | Included; one language pair per stream, any of 60 languages into any other | Add-on $0.65/hr; up to 5 targets; English-paired (~34 languages); not available on Agent STT |
| Speaker labels | speakers: true on the same stream | Included; default up to 50 speakers (can increase to 100) |
| Live minutes / summary | digest: true, pushed during the session | Summaries add-on $0.12/hr listed for pre-recorded; not a live digest on the same stream |
| Latency (vendor docs) | ~0.28 s to first word; translation +~88 ms | Partials under ~500 ms; finals roughly 0.7–2 s |
| Custom terms | Glossary + forced translation pairs, included | Custom dictionary included on Standard / Enhanced / Linden |
| Concurrent streams | Not published on the developers page | 2 on Free, 50 on Pro self-serve; Enterprise scales further |
| Long sessions | Connect token limits time to open; server backlog capped at 2 s | Streams over 24 hours supported |
| Deployment | SaaS gateways in 5 regions | SaaS (EU / US / AUS processing); on-prem / on-device on Enterprise |
| Free allotment | None on the public rate | $100 signup credits, no card required |
| Compliance (public) | See LecSync terms / docs for current posture | SOC 2 Type II, ISO/IEC 27001:2022, GDPR, HIPAA listed |
| SLA | Best-effort, no SLA today | Enterprise / contract terms; self-serve is usage-based |
Sources: Speechmatics pricing, real-time product, translation feature, plans, LecSync connect docs.
What an hour, and 1,000 hours, actually cost
Figures below use Speechmatics pay-as-you-go list rates with no volume or model-training discount. Opting into model training takes 33% off Speech-to-Text rates; volume discounts start above 500 hours per month per product type.
| LecSync | Speechmatics Standard | Standard + translation | Enhanced + translation | |
|---|---|---|---|---|
| Captions | Included | $0.45/hr | $0.45/hr | $0.80/hr |
| Speaker labels | Included | Included | Included | Included |
| Translation | Included | — | $0.65/hr | $0.65/hr |
| Live minutes | Included | Not on live add-ons list | Not on live add-ons list | Not on live add-ons list |
| Per hour | $0.36 | $0.45 | $1.10 | $1.45 |
| 1,000 hours | $360 | $450 | $1,100 | $1,450 |
Linden 1 plus translation is $0.95 per hour on the same math, but Linden is positioned for voice agents with turn detection rather than general meeting captions. Melia 1 lists a realtime rate of $0.40 per hour on the pricing schema; translation is not offered on Melia in the add-on model list, and Melia realtime still shows as coming soon in parts of the product UI, so we keep Standard and Enhanced as the main comparison.
Captions alone already put Standard above LecSync's all-in price ($0.45 vs $0.36). Once you add translation, the gap widens to about 3× to 4×. For a wider look at add-on stacking across vendors, see what 1,000 hours cost with translation, speakers and summaries.
Where Speechmatics is the better pick
- You need on-prem or on-device. Containers, virtual appliances, hybrid and on-device options are Enterprise offerings. LecSync is SaaS gateways only.
- Compliance paperwork matters. Speechmatics publicly lists SOC 2 Type II, ISO/IEC 27001:2022, GDPR and HIPAA, plus a Trust Center.
- You want several translation targets at once. One Speechmatics job can request up to five English-paired target languages. LecSync takes one language pair per stream.
- Domain vocabulary is heavy. Custom dictionary is included on Standard, Enhanced and Linden; Enhanced is the accuracy-first SKU.
- You are building a voice agent. Linden 1 bundles turn detection and VAD at $0.30 per hour.
- You are testing. New accounts get $100 in credits with no card required; Free allows two concurrent realtime connections.
- Sessions run for a day or more. Speechmatics documents streams longer than 24 hours.
Where LecSync fits better
- Captions and translation on the same screen, any language pair. Translation streams alongside the transcript and adds about 88 ms to the first token. Any of 60 languages can be translated into any other, not only through English.
- Speakers and live minutes without extra jobs. Speaker labels come with timestamps, and live minutes build section by section while the session runs. See live minutes vs post-call summarization.
- One line on the invoice. $0.006 per accepted audio minute whether you enable one feature or all of them. No separate translation meter.
- Thin clients. A one-time token opens the socket, so a browser or mobile app never holds a long-lived ASR credential.
- Latency that does not drift after a stall. First words reach the screen in about 0.28 seconds, and the server caps backlog at 2 seconds.
The trade-offs are real. LecSync takes one language pair per stream, accepts only 16 kHz mono PCM in v1, does not publish a default concurrency quota, and offers no SLA today.
FAQ
Does Speechmatics realtime include translation in the base rate?
No. Translation is a separate add-on at $0.65 per hour on the public pricing page. It supports up to five target languages per job, with about 34 English-paired languages, and works in realtime for Standard, Enhanced and Linden. It is not available on Agent STT.
How much does Speechmatics Realtime Standard cost per hour?
The pay-as-you-go list rate is $0.45 per hour, billed to the second. Enhanced is $0.80 per hour. Speaker diarization and custom dictionary are included on those models; translation, summaries and other language-intel add-ons are priced separately.
Can Speechmatics label speakers in realtime?
Yes. Set diarization to speaker (or channel). The real-time product page says the default is up to 50 speakers, increasable to 100. Diarization is not a separate line item on the public add-ons list.
How many concurrent Speechmatics realtime streams can I run?
Free accounts get two concurrent realtime connections. Pro self-serve lists 50. Enterprise adjusts dynamically. Speechmatics also documents streams longer than 24 hours.
What does LecSync include in $0.006 per minute?
Realtime transcription in 60 languages, realtime translation, audio-aligned timestamps, speaker separation, glossary and domain context, AI translation refinement, live structured minutes, and optional 30-day storage of recordings and transcripts. The rate stays the same whichever of these you turn on.
Can Speechmatics translate between two non-English languages in one call?
The public translation language list is English-paired: you translate to or from English among the supported set. LecSync configures one language pair per stream and translates between any two of its 60 languages, with direction chosen per segment.
The short version
Choose by whether translation has to ride on the same live connection at a predictable minute price. For captions, any-to-any translation, speakers and live minutes on one socket at $0.36 per hour, LecSync is the simpler build. If you need on-prem deployment, multi-target English-paired translation, published compliance certifications, or a voice-agent model, Speechmatics is the stronger fit — at roughly $1.10 to $1.45 per hour once translation is on. Create a key in the LecSync developers console and follow the connect docs to open your first stream. If you are also weighing cloud vendors, compare Amazon Transcribe streaming and Azure AI Speech.