Accuracy, speed, and trust signals (with receipts)

The model stack

Transcription does not bet on a single vendor. ScreenApp routes each job through a provider chain tried in a fixed order, skipping any provider that does not support the requested language. xAI Grok Speech-to-Text leads the chain, followed by Mistral Voxtral, Modulate, Google Gemini, and Groq-hosted Whisper Large-v3 as the deepest fallback.

• xAI Grok Speech-to-Text: the default, first-choice provider. Highest published accuracy on phone-call and multi-channel audio (5.0% entity error rate vs ElevenLabs 12.0%, Deepgram 13.5%, AssemblyAI 21.3%). Native word-level speaker diarization in 25 languages. Built on the same production stack that powers Grok Voice inside Tesla vehicles and Starlink customer support. API priced at $0.10/hour batch, $0.20/hour streaming (roughly 60% below ElevenLabs and Deepgram per Dapta's comparison). Launched April 18, 2026.
• Google Gemini (gemini-2.5-flash): broad catch-all for audio outside the other providers' declared languages.
• OpenAI Whisper Large-v3 on Groq's inference infrastructure: the deepest fallback in the chain, used only once every provider above declines or is rate-limited. Broadest language coverage (99 languages), inherited from the fallback role rather than from being the default path. Large-v3 reduced word error rates 10 to 20% over Large-v2. Groq runs Whisper at a 164x real-time speed factor on the independent Artificial Analysis benchmark.

Mid-chain fallback providers (used when xAI declines or is rate-limited, before falling back further to Gemini and then Whisper): Mistral Voxtral and Modulate. A transcription job never fails because one vendor had an outage.

LLM layer for summarization, chat, and AI analysis: OpenAI GPT-5.6 Luna is the default model for AI chat, summaries, titles, templates and translation. Anthropic Claude Sonnet 5 and Google Gemini 3.1 Flash Lite can be selected for specific jobs instead. Watching video frames and listening to audio always runs on Google Gemini. Live captions use their own provider order, led by Groq-hosted Whisper for speed.

Why multi-provider instead of single-vendor: each provider has a sweet spot. xAI Grok STT has the lowest published error rate on phone calls, which is why it leads. Gemini is the broad catch-all and doubles as the downstream LLM. Whisper on Groq has the most languages, which is why it anchors the fallback chain rather than sitting idle. Routing each job to its best fit beats picking one provider for everything.

Customer audio is never used to train any of these models. Audio is processed and deleted per your account's retention settings. Full data handling on the Trust Center.

Live captions (beta)

Live captions use a separate provider order from batch transcription, led by Groq-hosted Whisper Large-v3 for latency rather than by xAI: Groq, then xAI, Mistral, Modulate, and Google Gemini. This order is chosen for speed during a live session, not for the accuracy fit that leads the batch chain. No accuracy or latency figures for live captions are published yet.

Speed: the Groq case study

This section is history, not a current benchmark. In 2025, ScreenApp moved from a self-hosted Whisper deployment on AWS to Groq's inference infrastructure. Groq published the case study below; the numbers are from their engineering team's measurements of that pipeline, which was the default at the time. Whisper Large-v3 on Groq is now the deepest fallback, not the default provider (xAI Grok Speech-to-Text leads the chain). No speed or latency figures have been published yet for the current xAI-led provider chain, or for live captions.

Source: ScreenApp + Groq case study (groq.com).

Historical, for the 2025 Groq Whisper pipeline: a 60-minute meeting completes in roughly 3 minutes end-to-end (transcription, diarization, summary generation). A 2-hour video processes in about 6 minutes. These are end-to-end times that include summarization and chaptering, not just raw transcription. No speed figures have been published yet for the current xAI-led chain.

YouTube caption import (not transcription)

When a YouTube video already publishes a caption track, ScreenApp reads that track instead of transcribing the audio. No speech recognition runs at all, which is why it is fast. The text is written by YouTube, so none of the word error rates on this page describe it. Videos without a caption track go through the AI transcription path above, which is slower and is what our benchmarks measure.

Method: both windows measure the same endpoint (import start to transcript indexed) on the same population, on production traffic. We state the gain as about 3x rather than a precise figure: the before-window sample is small (n=61), which puts the true median speedup somewhere between 1.9x and 3.7x at 95% confidence. The mean (3.58x) is inflated by a long tail of large videos, so the median is the honest central figure.

Two limits worth stating plainly. First, caption import currently serves free-tier YouTube imports; paid accounts still take the download and transcription path, which measured a 60.5 second median before and 44.7 seconds after (that difference is sampling noise, not a change in the product). Second, 14.7 seconds reflects the current implementation, which fetches captions with yt-dlp. Last refreshed: July 2026.

Accuracy: word error rate benchmarks

Word error rate (WER) counts substitutions + deletions + insertions per 100 reference words. Lower is better. Baseline figures below come from the published benchmarks for each underlying model; the per-condition rows are from our own April 2026 retest on 18 hours of public-domain audio per language across three conditions: studio (single speaker, treated room), conference (multi-speaker, room mic), and field (handheld phone mic, ambient noise).

Headline figure: 95.8% accuracy. This is measured on the April 2026 provider mix used for that retest, before xAI Grok Speech-to-Text became the default provider. It has not been re-measured on the current default provider.

Published baselines

• Whisper Large-v3: 2.7% WER on LibriSpeech test-clean (audiobook-quality audio); 8 to 12% WER on real-world English meetings, podcasts, and call recordings. Source: openai/whisper-large-v3 model card.
• xAI Grok Speech-to-Text: 5.0% error rate on phone-call entity recognition, compared to ElevenLabs at 12.0%, Deepgram at 13.5%, and AssemblyAI at 21.3%. This is xAI's own published benchmark, not a ScreenApp test. Source: xAI Grok STT launch announcement.

Per-language WER (April 2026 retest)

† iPhone mic WER is a projection, not a measurement. Computed as Field WER × 1.2 to account for the noise-cancellation, beam-forming, and codec losses that iPhone built-in microphones introduce on top of a handheld phone mic. It has not been measured on a handset; read this column as an estimate only until a measured run is published.

Speaker counts per language are small (2 to 4). The figures above are indicative, not statistically robust, and no confidence intervals are published for them.

Sample files are not published. The corpus is drawn from public-domain sources (see Test methodology above); customer audio is never used, and the individual audio files used for the retest are not posted publicly.

Test methodology

• Corpus: 18 hours per language, drawn from Common Voice contributions, public lecture archives, and journalist transcript releases. No customer audio is ever included.
• Scoring: Aligned with the original transcript using jiwer, the same library AssemblyAI references. Punctuation and capitalization are not penalized. Speaker labels scored separately.
• Cadence: Last full run: April 22, 2026. A retest on the current default provider chain (led by xAI Grok Speech-to-Text) has not been published yet. The numbers above are the April 22, 2026 results and will not change on this page until a newer full run is published here.
• Limitation: the provider fallback order has changed since the April 22, 2026 retest: xAI Grok Speech-to-Text is now the default provider and Whisper Large-v3 has moved to the deepest fallback position. The WER figures above were measured on the provider mix in place at retest time; they have not been re-run against the current default order. Treat this table as the last published measurement, not a guarantee of what the current default provider scores today, until a newer retest is published.
• Conditions defined: Studio = single speaker, lavalier or shotgun mic, treated room. Conference = multi-speaker, room mic, occasional overlap. Field = handheld phone mic, ambient crowd or traffic noise.

Speaker diarization

Diarization (attaching a speaker ID to each word) is a flag passed to whichever transcription provider handles the job, not a separate pipeline. It follows the same language-based provider order as transcription: xAI Grok Speech-to-Text, then Mistral Voxtral, then Modulate, then Google Gemini, then Groq-hosted Whisper Large-v3 as the deepest fallback. Routing is automatic; you do not pick.

• xAI Grok Speech-to-Text leads the chain and supports diarization. Word-level speaker IDs, with published coverage across 25 languages (xAI's figure). 5.0% entity error rate on xAI's own published phone-call benchmark vs ElevenLabs 12.0%, Deepgram 13.5%, AssemblyAI 21.3%, not a ScreenApp test. Launched April 18, 2026.
• Mistral Voxtral, Modulate, and Google Gemini also support diarization and are used in that order when xAI declines or is rate-limited. Groq-hosted Whisper Large-v3, the deepest fallback, does not diarize: a job that reaches Whisper is transcribed without per-speaker labels.

Whichever provider actually handles the job produces the diarized output in the same response as the transcript; there is no separate content-type router that picks "phone calls" versus "meetings." Speaker labels are available for the languages shown with a dagger in the language list below, up to 25 languages on xAI's published coverage.

• Modes: Real-time streaming (live captions, its own provider order) and batch (for uploads).

Sources: xAI Grok STT and TTS API announcement, Google Gemini.

Supported languages

ScreenApp supports 99 languages. That figure is the union across the whole transcription provider chain (xAI Grok Speech-to-Text, Mistral Voxtral, Modulate, Google Gemini, and Groq-hosted Whisper Large-v3), not a per-provider count: which languages actually route to which provider depends on which of them declare that language. Whisper Large-v3, the deepest fallback, contributes the broadest published language list on its own. A subset also supports word-level speaker diarization: those are the languages marked with † in the list below. xAI publishes diarization coverage for 25 languages in total, so the † marks here are the ones confirmed against our own list rather than the full xAI set. The list below currently runs one entry longer than the published 99 count (it includes Cantonese); we have not yet reconciled which number to publish.

Whisper's language list source: github.com/openai/whisper. Grok STT language coverage source: xAI Voice docs.

AI summaries and chat: what to expect

OpenAI GPT-5.6 Luna is the default model for AI chat, summaries, titles, templates and translation. Anthropic Claude Sonnet 5 and Google Gemini 3.1 Flash Lite can be selected for specific jobs instead. Watching video frames and listening to audio always runs on Google Gemini.

Summaries and chat answers are generated from the transcript, not from the original audio. A transcription error carries into the summary and into any answer that depends on that part of the transcript.

AI summaries and chat answers can be incomplete or wrong. For anything a decision depends on, check the summary or answer against the transcript or the original recording rather than taking it at face value.

Timestamps and the full transcript are there for exactly this: to let you jump to the source and verify what was actually said. No formal published evaluation of summary or chat quality exists yet.

Platform availability

ScreenApp is available as the web app, Mac and Windows apps, iOS and Android apps, and a Chrome extension that records tab, screen, mic and camera (Chrome, Edge, Brave, Opera and other Chromium browsers). Ratings and review counts below are pulled at build time from each platform's canonical store listing (the date is named next to each rating). Other listing details (version numbers, sizes, install dates) are point-in-time snapshots checked when this section was last edited; the live store listings are always the authoritative source.

iOS app (iPhone, iPad, Apple Silicon Mac via Catalyst)

• Listed name: ScreenApp: AI Voice Recorder
• App Store rating: 4.0 stars across 111 ratings on the US storefront (Apple iTunes Lookup, pulled September 15, 2026).
• Current version: 1.4.25, released March 23, 2026.
• First public release: March 18, 2025.
• Minimum iOS: 15.0.
• Installer size: 103.6 MB.
• Supported devices: universal binary covering iPhone 5s through iPhone 17 Pro Max, every iPad from iPad Air onward (including the M4 and M5 iPad Pro lines), iPod touch 6th and 7th generation, and Apple Silicon Macs via Mac Catalyst.
• Category and content rating: Productivity, age 4 plus, free.
• Publisher: ScreenApp Pty Ltd.
• App Store URL: apps.apple.com/us/app/screenapp-ai-voice-recorder/id6741723588.

iOS App Privacy nutrition label

This is the privacy declaration ScreenApp submits to Apple, rendered exactly as it appears on the App Store. Apple's listing is authoritative; if the table below ever diverges from the live App Store page, the App Store page wins and this table should be corrected. ScreenApp declares no tracking data.

Canonical source: the App Privacy section on the ScreenApp App Store page. Full data handling policies on the Trust Center.

macOS (native desktop app)

• Distribution: direct DMG download from screenapp.io/desktop. The Mac app is not currently on the Mac App Store; the DMG is the canonical install path.
• Installer size: 67.9 MB (ScreenApp-latest.dmg).
• Last build pushed: March 22, 2026.
• Versioning: rolling release. The installer URL is ScreenApp-latest.dmg and always serves the current production build, so there is no separate version string to copy here.
• Architectures: universal binary for Apple Silicon and Intel Macs.
• iOS app on Mac: Apple Silicon Macs can also run the iOS app above via Mac Catalyst from the App Store. The native DMG is recommended for desktop workflows that need system audio capture, larger uploads, or background recording.

Android

• Listed name: ScreenApp - AI Voice Recorder
• Google Play rating: 3.9 stars across 775 ratings (Google Play structured data, pulled September 15, 2026).
• Installs: 50,000 plus, as reported by the Google Play listing (Play reports installs in buckets, not an exact count).
• Last update: May 8, 2026.
• Category and content rating: Productivity, age 3 plus, free.
• Package name: io.screenapp.screenapp_mobile.
• Play Store URL: play.google.com/store/apps/details?id=io.screenapp.screenapp_mobile.

Web app

• Access: runs in the browser at screenapp.io/app, no install required.

Chrome extension

• Chrome extension that records tab, screen, mic and camera. Supported on Chrome, Edge, Brave, Opera and other Chromium browsers.

Rating and review counts move daily on the App Store and Google Play. The numbers on this page are point-in-time snapshots, dated above. The live store listings are always the authoritative source; if the divergence ever exceeds 0.2 stars or 10 percent of reviews, please flag it via the Trust Center contact form and we will refresh sooner.

Selected customer reviews

Two named customer reviews drawn from the public reviews page, one from Capterra and one from the Chrome Web Store. They are hand-picked samples, not a random draw and not the full corpus, so do not read them as an average. The aggregate ratings are whatever the App Store and Google Play report (4.0 stars across 111 ratings on iOS, 3.9 stars across 775 ratings on Android, pulled September 15, 2026), not the average of the two reviews below.

Production corpus

The numbers below are real production counts, pulled at build time from the same MongoDB cluster the dashboard reads from. They are not marketing rollups, not rounded, and not estimated. Refresh cadence: every deploy. Last pulled: September 15, 2026.

Recent activity (indexed proxy via videometainfo.createdAt): 1,810 recordings analysed in the last 24 hours, 8,000 in the last 7 days, 33,081 in the last 30 days. Daily rate of roughly 1,103 analyses per day.

Why videometainfo and not recordings directly: recordings._id is a UUID, so we cannot do indexed time-range queries on it. Each videometainfo doc maps 1:1 to a recording via the unique recordingId index, so the time-windowed counts above are a faithful proxy. Methodology and the open query module: below.

User base

8,291,249 accounts registered as of September 15, 2026. Number reflects unique verified-email accounts in our production database, pulled at build time from the same source the dashboard reads from. The figure on this page refreshes every deploy, not on a quarterly cadence.

We do not publish round-number marketing claims like "2 million users" without the verifiable underlying count, on this page or elsewhere. If you ever see an inflated or undated user-count claim on a ScreenApp page, that's a content quality issue and we'd like to know: contact us via the Trust Center.

How we count

Every numeric claim on this page and across screenapp.io that depends on production data follows the same pipeline. Numbers are not curated, edited, or rounded for marketing.

1. Source of truth: the ScreenApp production MongoDB cluster, the same database the dashboard, mobile apps, and backend services read from. No marketing database, no cached marketing CMS.
2. Build-time pull: the marketing site has no direct database access. At each deploy, the static-site build calls a read-only backend endpoint (GET /v2/site-data) that runs a small set of indexed aggregations and returns scalar counts. The query module is open inside the same repo at scripts/site-data-queries.ts.
3. Frozen for the build: the returned numbers are written to a local SQLite file (data/stats.db) and read synchronously by every page at static-generation time. Within a single deploy the numbers do not drift; between deploys they refresh.
4. Soft fail: if the endpoint is unavailable or returns an unexpected shape, the previous deploy's figures are reused and the build proceeds. The site never ships placeholder text in place of a missing number.
5. Last refresh: the data on this page was pulled on September 15, 2026.

If you ever spot a number on the site that disagrees with a figure on this page, please flag it via the Trust Center. A divergence is a bug.

Free access and pricing

Two ways to use ScreenApp without paying upfront:

• Free signup (Free Forever): Process one recording for free. No credit card. No expiry. After your first recording, you'll need a paid plan or the 7-day trial to keep going. This is what we mean when we say "Free Forever": the first recording is free indefinitely, not unlimited recordings free.
• 7-day Pro trial: Full access to the Pro plan for 7 days. Credit card required (we don't charge until day 8, and you can cancel anytime during the 7 days at no cost). After day 7, the card is billed $228/year.

Paid plans

• Pro: $19/month billed annually ($228/year), or $30/month billed monthly. Recordings up to 2 hours each. The 7-day trial above is for this plan.
• Max: $34/month billed annually ($408/year), or $69/month billed monthly. Adds unlimited AI chats, transcriptions and uploads, and recordings up to 3 hours. SSO and SAML are not available on any plan today.
• Monthly billing: Available at higher per-month rates without the 7-day trial.

Current pricing and feature breakdown on the pricing page. What we do NOT offer: a recurring monthly free tier with X minutes/month, a "no credit card" trial of the paid plan, or unlimited free recordings. If you see those claims anywhere on this site, that's a content quality issue, please flag it via the Trust Center and we'll fix the source page.

Security and compliance

SOC 2 Type 2 audited annually. 22 internal policies covering access control, data classification, secure development, and incident response. Continuous control monitoring.

• Encryption: AES-256 at rest, encrypted in transit.
• AI training: recordings are not used to train AI models.
• Retention after cancellation: recordings are kept when a paid plan lapses; free-plan limits apply.
• Subprocessors: the vetted subprocessor list is in the Privacy Policy.

Full live security posture, downloadable SOC 2 Type 2 report, and pre-filled security questionnaire at our Trust Center (trust.inc/screenapp).

Sources and external benchmarks

• ScreenApp + Groq case study: speed and cost numbers for the inference layer.
• Groq: Whisper at 164x real-time: independent Artificial Analysis benchmark of Whisper Large-v3 on Groq inference.
• xAI Grok Speech-to-Text launch: diarization benchmarks, language list, Tesla and Starlink production references.
• Dapta: Grok Voice API pricing comparison: independent confirmation that Grok STT undercuts Deepgram and ElevenLabs by roughly 60%.
• MarkTechPost: xAI launches Grok STT and TTS APIs: third-party coverage of the launch.
• OpenAI Whisper Large-v3 model card: published WER baselines and the 10 to 20% improvement over Large-v2.
• openai/whisper on GitHub: full 99-language list and model details.
• Artificial Analysis Speech-to-Text leaderboard: independent third-party WER and speed benchmarks across providers.
• jiwer: the WER scoring library used in our per-language retests.
• ScreenApp Trust Center: SOC 2 Type 2 report, 22 internal policies, sub-processors, security questionnaire.

Errata and corrections

Numbers on ScreenApp pages should match this page. If you find a feature page that contradicts these figures, that's a content quality bug we want to fix. Report it via the Trust Center contact form so we can check and correct the source page.

Corrections log

• 2026-09-15: corrected four things on this page. The default AI model is OpenAI GPT-5.6 Luna (this page previously said Gemini end-to-end and "not powered by GPT or Claude"). The transcription provider order changed: xAI Grok Speech-to-Text is now first, not Whisper on Groq. The retest cadence promise ("retested quarterly," a fixed next-run date) was removed because no retest against the current provider chain has been published. The speaker diarization section was corrected: it described content-type routing (xAI for phone calls, Gemini for meetings on a shared Vertex AI pipeline) that does not match the code; diarization actually follows the same language-based provider order as transcription.