LecSync vs Rev AI Streaming: Live Translation, Speakers and Real Cost

Rev AI's streaming API is one of the cheapest ways to get English captions on screen. Rev's pricing page lists Reverb transcription at $0.20 per hour. What the live socket does not do is translate: translation and summarization are sold as Insights on async jobs, and the streaming docs list nine languages. The LecSync Realtime Speech API returns captions, a translation track, speaker labels and live minutes on one WebSocket for $0.006 per audio minute.

Everything below comes from public pages. Rev AI figures are from the pricing page, the streaming API overview, the streaming request reference, streaming billing and the FAQ, all read on October 12, 2026. LecSync figures are from the developers page and the pricing docs.

The short answer

If your product only needs live captions in English, or in one of Rev AI's nine streaming languages, Rev AI costs less: $0.20 per hour for English and $0.30 per hour for other languages, against LecSync's $0.36 per hour (60 × $0.006).

If you need a second language on screen while people are still talking, speaker labels on every final segment, or minutes that build during the session, LecSync returns all of it on the same connection at the same $0.36 per hour. On Rev AI, translation and summaries are not part of the streaming API; they run on async jobs after the audio is complete.

What each socket returns

With LecSync, your server calls POST /v1/realtime/connect and gets back a stream URL and a one-time token. The client opens the WebSocket and pushes 16 kHz mono PCM. Transcript segments, their translations (paired by segment ID), speaker labels with timestamps (speakers: true) and live minutes (digest: true) all come back on that socket.

With Rev AI, you open wss://api.rev.ai/speechtotext/v1/stream with an access token and a content type, then receive partial and final hypotheses. A few details matter when you plan the integration:

• Speakers. enable_speaker_switch=true works only with transcriber=machine_v2. It adds a speaker_id to final hypotheses, and the docs say that number is incremented each time the active speaker changes.
• Vocabulary. The language parameter cannot be combined with custom_vocabulary_id, so custom vocabulary applies to the default English stream.
• Translation. Rev's FAQ points to translation_config on async job submission. The streaming request reference has no translation parameter.
• Browser clients. Short-lived streaming tokens exist, but only in the US region and not yet in Rev's SDKs.

If you were planning to bolt a separate MT service onto a caption stream, we compared that route in built-in translation vs a DIY STT + MT pipeline.

Side-by-side facts (checked October 12, 2026)

What you pay: per hour, per 1,000 hours, and on short streams

All Rev AI numbers below are pay-as-you-go list prices with no Enterprise discount.

The last column assumes you resubmit the recording as an async English job with standard translation ($0.002 per minute). With premium translation ($0.025 per minute) it becomes $1.90 per hour. In both cases the translation arrives after the meeting, not on screen during it.

Billing rules matter too, especially for short or bursty audio. Rev AI bills each stream for the larger of wall-clock time and audio sent, with a 15-second minimum, and holds 10 minutes of credits when the socket opens. LecSync bills accepted audio down to fractions of a second, and connections that never send audio cost nothing. Take 10,000 voice commands of five seconds each, one stream per command:

• Rev AI: 10,000 × 15 s = 41.7 hours × $0.20 = $8.33
• LecSync: 10,000 × 5 s = 833 minutes × $0.006 = $5.00

For long, continuous English captioning, Rev AI's lower hourly rate wins. For a wider look at how add-ons stack across vendors, see what 1,000 hours cost with translation, speakers and summaries.

Where Rev AI is the better pick

• English-only captions on a tight budget. $0.20 per hour is well below LecSync's all-in $0.36.
• Your audio already arrives as RTMP or in mixed formats. Rev AI takes RTMP ingest and any FFmpeg format. LecSync v1 accepts 16 kHz mono PCM only.
• Healthcare work. HIPAA-compliant processing is available once enabled on the account (RTMP is excluded under HIPAA).
• EU data residency. On request, Rev AI processes and stores data in Frankfurt, with a separate EU streaming endpoint.
• You also want a polished transcript afterwards. The same vendor offers async transcription in 58+ languages and human transcription at $1.99 per minute.
• You want to test first. New accounts get credits worth 5 hours of Reverb, and the default concurrency (10 streams) is published.

Where LecSync fits better

• A second language on screen during the session. Any of 60 languages can be translated into any other. Translation streams with the transcript and adds about 88 ms to the first token.
• Languages beyond Rev's streaming nine. LecSync's live stream covers 60.
• Speaker labels in every language. Every final segment carries a speaker label with timestamps, so a person's turns can be found in the recording. See building a meeting client with speakers and timestamps.
• Minutes while the meeting runs. digest: true pushes a section of minutes over the same connection once enough has been said.
• Terminology that holds in both languages. The glossary biases recognition and fixes how names and product terms are translated.
• Short, bursty audio. No 15-second floor and no billing for wall-clock idle time.
• Thin clients. The one-time token from connect opens the socket, so a browser or mobile app never holds a long-lived key.
• Latency that does not drift. The server caps backlog at 2 seconds after a network stall instead of queueing it.

LecSync's limits are real: one language pair per stream, 16 kHz mono PCM only in v1, no published concurrency quota or maximum session length, no free allotment, and no SLA today.

FAQ

Does Rev AI streaming support real-time translation?

Not on the stream. The streaming request reference has no translation parameter, and Rev's FAQ points to translation_config on async jobs. Rev lists translation in 11 languages at $0.002 per minute (standard) or $0.025 per minute (premium), applied after transcription.

How much does Rev AI streaming cost per hour?

Rev's pricing page lists Reverb transcription at $0.20 per hour for English and $0.30 per hour for other languages, and the streaming docs point non-contract customers to that page. Each stream is billed for the larger of wall-clock time and audio sent, rounded up to the second, with a 15-second minimum.

Which languages does Rev AI streaming support?

Nine: English, French, German, Italian, Japanese, Korean, Mandarin, Portuguese and Spanish. The async API covers 58+ languages. Some streaming options, such as skip_postprocessing and priority, work only for English and Spanish.

Can Rev AI label speakers in a live stream?

Partly. With transcriber=machine_v2 and enable_speaker_switch=true, final hypotheses include a speaker_id that is incremented when the active speaker changes. Rev's FAQ says diarization supports up to 8 speakers in English and 6 in other languages, and that speaker identification is not supported.

How many Rev AI streams can run at once, and for how long?

The default concurrency limit is 10 streams, and support can raise it. Each stream can run for up to 3 hours; before that, you open a new WebSocket and switch audio to it once it reports connected.

What does LecSync include in $0.006 per minute?

Realtime transcription in 60 languages, realtime translation between any two of them, aligned timestamps, speaker labels, glossary and domain context, AI translation polish, live structured minutes, and optional 30-day storage of the recording and transcript. Turning features on or off does not change the rate.

Bottom line

Choose Rev AI when you need cheap, long-running English captions, RTMP ingest, HIPAA processing or EU residency, and translation can wait until after the session. Choose LecSync when captions, a live translation track, speaker labels and running minutes have to arrive together during the session, at one flat $0.36 per hour. Create a key on the LecSync developers page and open your first stream with the connect docs. Still comparing? See LecSync vs Speechmatics and LecSync vs Deepgram.