Best Voice to Text Apps for iPhone in 2026
Lena Vollmer
Disclosure: This guide is written by the Aqua Voice team. We've done our best to evaluate each app fairly, but you should know we have skin in the game. All benchmark numbers are cited where possible so you can verify them independently.
Typing on glass is slow. Even the fastest iPhone typists top out around 40-50 words per minute. Speaking is 3-4x faster, and the voice-to-text apps available on iPhone in 2026 are dramatically better than they were even a year ago.
But "better" varies wildly depending on what you need. Some apps are designed for transcribing recorded audio. Others replace your keyboard system-wide. Some use Apple's built-in speech engine under the hood. Others run proprietary models that handle technical vocabulary, multiple languages, or professional jargon.
We tested the most popular options across accuracy, speed, language support, and how well they handle real-world use (not just reading scripted sentences into a microphone). Here's what we found.
Quick Comparison
| App | Best For | Accuracy | Speed | Price | Works System-Wide | Offline |
|-----|----------|----------|-------|-------|-------------------|
| Aqua Voice | Professionals, developers, multilingual | 94.5% (5.52% WER) | Sub-second | $10/mo | Yes (iOS keyboard) | No | | Apple Dictation | Quick, casual messages | ~82% | Instant (on-device) | Free | Yes (built-in) | Yes | | Otter.ai | Meeting transcription | ~88% | Near real-time | Free / $17/mo | No (in-app only) | No | | Whisper Transcription | Offline transcription | ~92% | Batch (not real-time) | $5 one-time | No (in-app only) | Yes | | Audionotes | Voice memos to notes | ~85% | Near real-time | Free / $10/mo | No (in-app only) | No | | Rev | Professional transcription | ~90% | Batch | $0.25/min | No (in-app only) | No | | Gboard Voice | Android converts, casual use | ~84% | Instant | Free | Yes (keyboard) | Partial |
1. Aqua Voice
Aqua Voice launched on iPhone on March 1, 2026, bringing its Avalon speech model from desktop to iOS. The app works as a system-wide iOS keyboard, meaning you can use it anywhere you type: Messages, Slack, email, notes, browsers, anything.
What makes it different: Aqua Voice runs a proprietary model called Avalon, not Apple's speech engine or OpenAI's Whisper. The practical difference shows up in technical vocabulary, proper nouns, and non-English languages. Say "refactor the useState hook in the auth-utils package" and it comes through correctly, hyphens and camelCase included. Say the same thing to Apple Dictation and you'll get "use state hook in the off utils package."
Accuracy: 5.52% word error rate on general speech (LibriSpeech benchmark). 97.4% accuracy on technical/AI terminology (compared to 65% for Whisper-based tools on Aqua's internal developer-speech evaluation set).
Speed: Sub-second end-of-speech-to-text latency. Text appears almost immediately after you stop speaking. There's no waiting, no spinning indicator.
Languages: Full optimized support for English, Japanese, Spanish, German, French, and Russian. Whisper-level support for 49 languages total. Japanese users in particular benefit because voice bypasses the IME typing pipeline (romaji to hiragana to kanji selection) entirely.
Cleanup: Aqua Voice automatically removes stutters, false starts, and filler words (um, uh, "like"). It formats text appropriately for the app you're using. You speak naturally; clean text comes out.
Price: $10/month. If you already have a desktop subscription, iPhone is included at no extra cost. Free trial available (1,000 words, no credit card).
Limitations: Requires internet connection (the model runs in the cloud). No offline mode. Not designed for transcribing recorded audio files, only live dictation.
Best for: Anyone who uses voice input as a daily tool, not just for the occasional hands-free text. Developers, writers, professionals who need accuracy on specialized vocabulary. Multilingual users, especially Japanese.
2. Apple Dictation (Built-In)
Every iPhone has voice-to-text built in. Tap the microphone icon on the keyboard, speak, and text appears. It's free, it's always there, and for simple messages it works fine.
Accuracy: Solid for everyday language. Struggles with proper nouns, technical terms, and anything outside common vocabulary. "Schedule a meeting with Satya Nadella about the Kubernetes migration" will likely produce at least one error. Apple's on-device model prioritizes speed and privacy over specialized accuracy.
Speed: Near-instant. The on-device processing means there's no network latency. Text appears as you speak.
Languages: Supports 60+ languages with on-device processing for many of them. Language switching is automatic in some cases.
Cleanup: Basic punctuation insertion (periods, question marks). No stumble removal. If you say "um" or restart a sentence, it transcribes that literally.
Price: Free.
Limitations: No stumble cleanup. Limited accuracy on technical vocabulary. Stops listening after pauses (frustrating for longer dictation). No custom vocabulary or context awareness.
Best for: Quick texts, simple notes, hands-free replies. If your dictation needs are casual and infrequent, Apple's built-in option is good enough and you already have it.
3. Otter.ai
Otter is a meeting transcription tool, not a keyboard replacement. It records audio and produces transcripts, often in real-time during meetings.
Accuracy: Strong for conversational English, especially in meeting contexts. Speaker identification is a standout feature. Less reliable with technical jargon or non-English speech.
Speed: Real-time transcription during meetings. For uploaded audio, processing takes a few minutes.
Languages: English-focused. Limited multilingual support compared to dedicated voice input tools.
Features: Speaker identification, automated summaries, action item extraction, integrations with Zoom and Google Meet. The AI summary feature is genuinely useful for catching up on meetings you missed.
Price: Free tier (300 minutes/month). Pro at $17/month. Business plans available.
Limitations: Not a keyboard replacement. You can't use Otter to dictate into other apps. It's a standalone recording and transcription tool. The free tier has meaningful limits.
Best for: Meeting transcription and note-taking. If you sit in a lot of calls and want searchable transcripts with speaker labels, Otter does this well.
4. Whisper Transcription
Several iOS apps wrap OpenAI's Whisper model for on-device transcription. "Whisper Transcription" by developer Jordi Bruin is one of the most popular. These apps run the model locally on your iPhone, meaning they work offline.
Accuracy: Whisper large-v3 achieves roughly 7.44% WER on general benchmarks. In practice, the on-device versions use smaller, quantized models that are somewhat less accurate but still strong for general speech.
Speed: Not real-time. You record audio, then the app processes it. Processing time depends on the model size and your iPhone's chip. On newer iPhones (A17+), a 5-minute recording processes in under a minute.
Languages: Whisper supports 99 languages. Coverage and accuracy vary by language.
Price: Most Whisper-based apps are one-time purchases ($3-7). No subscription.
Limitations: Batch processing only, not live dictation. You can't use it as a keyboard. Technical and domain-specific accuracy is limited (65% on developer terminology in benchmarks). Smaller on-device models sacrifice some accuracy for speed.
Best for: Offline transcription of recordings. If you record voice memos or interviews and want accurate-enough transcripts without sending audio to the cloud, Whisper-based apps are a solid choice.
5. Audionotes
Audionotes is a voice-first note-taking app. You speak, it transcribes, and then uses AI to organize and summarize your notes.
Accuracy: Uses a combination of speech engines (varies by plan). Adequate for general note-taking, not specialized for technical vocabulary.
Speed: Near real-time transcription.
Features: AI-powered organization, tagging, and summarization. Can turn rambling voice notes into structured text. Integrates with some productivity tools.
Price: Free tier with limits. Premium at $10/month.
Limitations: Not a system-wide keyboard. Works within its own app only. The AI organization features are the differentiator, not raw transcription quality.
Best for: People who think out loud and want their scattered voice notes turned into organized, actionable text. More of a "second brain" tool than a voice input tool.
6. Rev
Rev offers both AI transcription and human transcription services via their iOS app.
Accuracy: AI transcription is solid (~90%+). Human transcription is 99%+ but costs more and takes longer.
Speed: AI transcription is near real-time for short recordings. Human transcription takes hours.
Price: AI transcription at $0.25/minute. Human transcription at $1.50/minute.
Limitations: Not a keyboard replacement. Designed for transcribing recordings, not live dictation. Per-minute pricing adds up for heavy users.
Best for: Professional transcription where accuracy is critical (legal, medical, journalism). The human transcription option is valuable when you absolutely cannot have errors.
7. Gboard Voice Typing
Google's keyboard app for iOS includes voice typing powered by Google's speech recognition.
Accuracy: Comparable to Apple Dictation for everyday speech. Google's model handles some proper nouns better (especially Google-ecosystem terms). Technical vocabulary is still a weakness.
Speed: Near-instant, similar to Apple Dictation.
Languages: Excellent multilingual support. Google's speech models cover 100+ languages.
Price: Free.
Limitations: Requires switching to Gboard as your keyboard (some people dislike the layout or privacy implications). Voice accuracy is good but not specialized. No stumble cleanup.
Best for: Users who already prefer Gboard for its other features (swipe typing, GIF search, translate) and want decent voice typing bundled in.
How We Tested
We evaluated each app on:
Accuracy: Dictated the same set of 50 sentences covering everyday language, technical terms, proper nouns, and mixed-language phrases. Counted errors.
Speed: Measured time from end of speech to final text appearing.
Real-world use: Used each app as a daily driver for a week, noting frustrations, failure modes, and standout moments.
Language handling: Tested English, Japanese, and Spanish dictation across all apps that claimed multilingual support.
The Bottom Line
If you just need occasional voice-to-text for quick messages, Apple Dictation is built in and free. Use it.
If you transcribe meetings, Otter is the best option with speaker identification and summaries.
If you want voice input as a real, daily keyboard replacement that handles technical vocabulary, multiple languages, and natural speech patterns, Aqua Voice is the clear choice. The accuracy gap between Aqua Voice's Avalon model and everything else on this list is significant, especially for professional use.
The iPhone app is new (launched March 1, 2026), but the underlying model has been trusted by 20,000+ paying subscribers on desktop. Same engine, new platform.
Try Aqua Voice free: aquavoice.com/download
