Aqua Voice vs Willow Voice summary
Aqua Voice and Willow Voice are both voice dictation apps for Mac, Windows, and iPhone, each built by a Y Combinator startup in the last two years. The core difference is the speech engine. Aqua runs its own proprietary model, Avalon, which debuted at #1 among proprietary models on the independent OpenASR leaderboard in October 2025. Willow does not publicly disclose what speech recognition powers its product. Aqua is $8/month (billed annually) and tuned for technical and coding speech; Willow is $15/month with a larger weekly free tier.
The quick verdict
Pick Aqua Voice if:
You dictate code, AI prompts, or technical jargon (Avalon is purpose-trained for this)
You want a proprietary speech model with independent third-party benchmarks
You're price-sensitive ($8/month vs $15)
You're switching from Apple Dictation or Dragon and accuracy is the reason
Pick Willow if:
You need 100+ languages (Aqua supports 49)
You need a verified HIPAA BAA (Aqua does not publish one)
You want a larger free tier for casual use (Willow: 2,000 words/week vs Aqua: 1,000 lifetime)
You need an offline mode on Mac or iOS
Aqua Voice vs Willow Voice: Feature-by-feature comparison
Aqua Voice
Willow Voice
Speech recognition model
Avalon (own, proprietary)
Not publicly disclosed
Independent benchmark
OpenASR: #1 proprietary at Oct-2025 debut
Not listed
AISpeak (AI/coding terms)
97.3%
Not published
9to5Mac vs Apple Dictation
1 error
Not tested
Mac / Windows / iPhone
✅ / ✅ / ✅
✅ / ✅ / ✅
Android
❌
❌
Languages
49
100+
Writing-style personalization
✅
✅
Context awareness (reads active app)
✅
✅
Custom dictionary
✅ (800 on Pro)
✅
SOC 2 Type II
✅
✅
HIPAA Ready
❌ (no public BAA)
✅ (per Willow)
Zero Data Retention
✅ (Enterprise)
Not disclosed
SSO / SAML / SCIM
✅ (Enterprise)
✅ (Enterprise)
Offline mode
❌ (cloud)
✅ (Mac & iOS)
Developer API
✅ Avalon API, $0.39/hr
Not public
Free tier
1,000 lifetime words
2,000 words / week
Pro price
$8/mo (billed annually)
$15/mo
Team plan
$12/seat/mo
$10/seat/mo (min 3)
Whose model is it?
Aqua trains and operates its own speech recognition model, called Avalon. Avalon 1 launched in August 2025, and at its debut on the OpenASR leaderboard (an independent benchmark maintained by Hugging Face) in October 2025, it ranked #6 overall and #1 among proprietary models. Avalon 1.5 (April 2026) reduced the average word error rate to 5.55%. Willow Voice does not publicly disclose what speech recognition technology powers its product.
This is the most defensible difference between the two products. If the model matters to you, one of these companies publishes independent benchmarks and the other does not.
How accurate is it on technical and coding speech?
Most dictation apps run on speech models trained on audiobook and meeting-style speech. They mangle technical vocabulary like "kubectl," "GPT-4o," "tsconfig," and "zshrc." Aqua trained Avalon specifically on human-computer interaction speech, including code, AI prompts, and developer email. On AISpeak, Aqua's published benchmark, Avalon scores 97.3%, versus Whisper Large v3 at 65.1% and ElevenLabs Scribe at 78.8%. Willow does not publish accuracy benchmarks for technical speech.
Is it independently tested?
9to5Mac independently tested Aqua against Apple's built-in macOS Dictation in August 2025, transcribing Steve Jobs' 2005 Stanford commencement address. Aqua produced 1 error. Apple Dictation produced 17. Aqua is also listed on the OpenASR Leaderboard, a public Hugging Face benchmark. Willow has been favorably covered by the tech press but, as of this comparison, has not been tested in a head-to-head dictation accuracy benchmark.
Which is better for Cursor, Claude Code, and AI prompting?
Aqua is built around the use case of talking to AI tools. Avalon is trained on AI-prompt speech, the desktop app reads the active application to adjust tone (casual in Slack, code-aware in Cursor), and the Avalon API ships at $0.39 per audio hour for developers building voice into their own apps. Willow also supports dictation into Cursor and other tools, but without published accuracy data on technical terms.
Pricing
Aqua Pro costs $8 per month, billed annually ($96/year), with a monthly option available. Willow Pro costs $15 per month, so Aqua is roughly 47% cheaper for individuals. Aqua's Team plan is $12 per seat per month; Willow's is $10 per seat per month with a three-seat minimum. Both offer enterprise plans with SSO and centralized billing. Willow offers a larger free tier (2,000 words per week versus Aqua's 1,000 lifetime words), and students can get Aqua at 70% off with a .edu email.
Privacy and data handling
Both Aqua and Willow are SOC 2 Type II certified. Aqua's Enterprise tier includes Zero Data Retention; Willow's HIPAA Ready status makes it the better fit for regulated healthcare workflows. Both are cloud-based, and Willow additionally offers an offline mode on Mac and iOS. Aqua's plans include a Privacy Mode, with organization-wide enforcement available on Team and Enterprise.
Tenure and pedigree
Aqua Voice joined Y Combinator's W24 batch in early 2024. Willow launched in March 2025 as part of YC's X25 batch, so Aqua has the longer operating history in voice dictation and a shipped, independently benchmarked model.
How to decide
Try both. They're both free to start. If you work primarily with AI tools or write code and care about accuracy, you'll notice the difference quickly. If you need a language Aqua doesn't cover, a HIPAA BAA, or an offline mode, Willow may be the better fit. Spend 10 minutes with each, dictating what you actually type day to day. The right choice will be obvious.
Frequently asked questions
What's the main difference between Aqua Voice and Willow Voice?
Aqua Voice runs on its own proprietary speech recognition model, Avalon, which is benchmarked publicly on the independent OpenASR leaderboard, where it debuted at #1 among proprietary models in October 2025. Willow Voice does not publicly disclose what speech recognition powers its product. Aqua is also $8/month (billed annually) versus Willow's $15/month.
Which is more accurate?
On published benchmarks, Aqua's Avalon scores a 5.55% average word error rate on OpenASR (Avalon 1.5, April 2026) and 97.3% on AISpeak, Aqua's public benchmark for AI and coding speech. Willow does not publish independent accuracy data. In a 9to5Mac side-by-side test against Apple Dictation, Aqua produced 1 error versus Apple's 17 on the same passage.
Which is cheaper?
Aqua Pro is $8/month billed annually ($96/year). Willow Pro is $15/month. Aqua is roughly 47% cheaper for individual users. Both offer free tiers and student discounts.
Does Aqua Voice have an iPhone app?
Yes. Aqua launched on iOS as an AI voice keyboard in April 2026, so you can dictate directly into any app. Willow also runs on iOS. Neither offers an Android app.
Does Aqua Voice have a team or enterprise plan?
Yes. Aqua's Team plan is $12 per seat per month with centralized billing and org-wide Privacy Mode. The Enterprise tier adds SSO/SAML, SCIM, Zero Data Retention, and shared team dictionaries.
Is Aqua Voice SOC 2 compliant?
Yes. Aqua is SOC 2 Type II certified, with a public Vanta trust center. Willow is also SOC 2 Type II certified.
Which has better language support?
Willow supports 100+ languages. Aqua supports 49.
Which has HIPAA support?
Willow advertises HIPAA Ready status, which makes it the better fit for regulated healthcare workflows. Aqua does not publish a HIPAA BAA.
Which is better for developers building voice into their own apps?
Aqua offers the Avalon API, an OpenAI-compatible speech-to-text endpoint at $0.39 per hour of audio. Willow does not publicly offer a developer API.