Relying on memory or manual typing during multi-person meetings leads to missed action items, but navigating free-plan limitations and hidden paywalls in meeting software is equally exhausting.
Free AI note takers in 2026 fall into three distinct structural categories: Cloud Meeting Bots (Otter, Fireflies, tl;dv, Notion AI), Native System Transcribers (Apple Voice Memos, Google Recorder, Windows Sound Recorder), and Open-Source Local Utilities (local Whisper GUIs). Each category solves specific transcription needs, but every free option imposes hard caps on monthly minutes, export permissions, speaker identification, or environment access.
This guide breaks down 8 free AI note takers across all three categories, details their operational limits, analyzes the hidden workflow costs of zero-budget tools, and provides a structural comparison matrix to help you choose the right setup.
Category 1: Cloud and Web AI Meeting Bots
Cloud meeting bots join your scheduled video calls as virtual participants, capturing audio directly from platforms like Zoom, Google Meet, and Microsoft Teams. They offer the most integrated experience—calendar syncing, automated join, and post-meeting summaries—but every free tier has a ceiling.
1. Otter.ai (Basic Free Tier)
Otter remains the most recognized name in AI meeting transcription, and its free tier provides genuinely useful functionality for light users.
- Free Quota & Limits: 300 total transcription minutes per month with a hard cap of 30 minutes per conversation[1]. File imports are restricted to 3 lifetime audio/video uploads.
- Speaker Identification: Supported on the free tier, with basic speaker labeling and auto-detection.
- In-Person Support: Supported via the mobile app's live microphone recording, though it requires an active internet connection for real-time cloud processing.
- Export Restrictions: Basic text export is allowed; advanced exports including PDF, SRT, and audio downloads are locked behind paid plans.
Where it caps out: If your team runs 45-minute client calls, Otter Basic stops transcribing at the 30-minute mark. The meeting continues, but the remaining 15 minutes of discussion, decisions, and action items vanish from the transcript. For users with more than 10 standard meetings per month, the 300-minute pool depletes within the first two weeks.
2. Fireflies.ai (Free Plan)
Fireflies takes a different approach to free-tier economics: instead of capping transcription minutes, it restricts storage and export access.
- Free Quota & Limits: Unlimited transcription minutes for live web meetings across 100+ languages, but team recording storage is hard-capped at 400 total minutes[2]. Once storage fills, older transcripts must be manually deleted or the account upgraded. Additionally, the free plan includes 20 AI credits per month.
- Speaker Identification: Supported across all supported languages.
- In-Person Support: Limited; the platform's primary design centers on web calendar bots joining Zoom, Teams, and Google Meet calls.
- Export Restrictions: Downloads and exports of raw transcripts or summaries are strictly blocked on the free tier. Users can view notes within the web workspace but cannot extract them.
Where it caps out: Fireflies gives you transcription without minute limits, then walls off the output. If your workflow requires downloading transcripts for compliance, sharing with absent team members, or feeding notes into a CRM, the free plan functions as a read-only archive that fills up at 400 minutes. For sales teams logging call notes into a pipeline, this creates a manual copy-paste bottleneck.
3. tl;dv (Free Forever Plan)
tl;dv positions itself as the generous option with "unlimited" recording, but the free tier's AI capabilities deplete rapidly.
- Free Quota & Limits: Unlimited meeting recording and transcription with a 3-hour limit per individual meeting[3]. Automated AI meeting summaries are capped at 10 lifetime uses on the free tier; after exhausting those credits, only the first 10 minutes of any call receive AI summarization.
- Speaker Identification: Not supported on the free plan. Transcripts appear as continuous text blocks without distinct speaker tags.
- In-Person Support: No; the tool requires integration with Zoom, Google Meet, or Microsoft Teams web calls.
- Export Restrictions: Direct exports to CRM or project management tools are paywalled; manual text copying is permitted.
Where it caps out: Ten AI summaries across a user's entire lifetime with the tool. A freelancer taking three client calls per week will exhaust this allocation within the first month. Beyond that, tl;dv records everything but asks the user to manually comb through full transcripts without speaker labels—essentially reverting to a raw recording tool.
4. Notion AI Meeting Notes (Free Trial Tier)
Notion AI is not a permanently free meeting transcription service, despite what trial-period roundups sometimes suggest.
- Free Quota & Limits: Free and Plus workspace accounts receive a small trial credit pool of approximately 20 AI responses or summaries[4]. Ongoing AI meeting transcription requires upgrading to the Business tier at $20 per user per month billed annually.
- Speaker Identification: Not supported; the feature processes single-stream desktop system audio or uploaded files.
- In-Person Support: Can transcribe local audio files imported into a workspace page, provided an internet connection is available.
- Export Restrictions: Notes reside natively inside Notion blocks and can be exported as Markdown, HTML, or PDF.
Where it caps out: Notion AI Meeting Notes functions as an extended free trial, not a free tool. Once the 20-response pool is consumed, the feature locks entirely. For teams already paying for Notion, the Business upgrade may be justifiable; for individuals searching for a zero-budget option, this tool stops working after roughly two weeks of regular meeting use.
Category 2: Native Operating System AI Transcribers
Built into phones and desktops, these tools process audio locally using on-device machine learning hardware. They eliminate monthly minute caps and cloud privacy concerns entirely, but trade away the automated summarization and calendar integration that cloud bots provide.
5. Apple Voice Memos & Notes Transcription (iOS 18 / macOS Sequoia)
With iOS 18 and macOS Sequoia, Apple introduced on-device speech-to-text transcription directly into Voice Memos and Notes, powered by the Apple Neural Engine.
- Free Quota & Limits: 100% free with no monthly minute caps or duration limits[5]. All processing occurs on-device on iPhone 12 and newer devices.
- Speaker Identification: Not supported in the initial release. Transcripts generate as continuous text blocks with no speaker attribution.
- In-Person Support: Full support; records offline in-person meetings, lectures, and voice notes directly through device hardware.
- Export Restrictions: No restrictions; users can copy text directly or export as Apple Notes and text documents.
Where it caps out: A lecture, interview, or meeting captured with Apple Voice Memos produces a raw, unattributed text wall. For solo dictation or single-speaker content, this is sufficient. For any conversation involving two or more people, the lack of speaker diarization means the user must manually parse who said what—a time cost that grows with meeting length and participant count.
6. Google Recorder (Pixel 3+)
Google Recorder represents the most capable free on-device option for multi-speaker transcription, but its availability is hardware-gated.
- Free Quota & Limits: 100% free with unlimited recording duration and no monthly caps on supported Pixel hardware. Supports continuous recordings up to 18 hours per file.
- Speaker Identification: Fully supported in real time. On Pixel 6 and newer devices, Google's Turn-to-Diarize model labels speakers as Speaker 1, Speaker 2, and so on[6] during live transcription in US English.
- In-Person Support: Excellent; designed specifically for offline, in-person voice recording, interviews, and lectures.
- Export Restrictions: Unrestricted; direct one-tap export to Google Docs, TXT, or WAV audio files.
Where it caps out: The hardware requirement is the ceiling. Anyone not using a Pixel phone cannot access Google Recorder. Additionally, the speaker diarization capability is currently optimized for US English, meaning users recording conversations in other languages receive transcripts without speaker labels. For Pixel owners conducting English-language interviews, however, this is the best free option available.
7. Windows Sound Recorder (Windows 11)
Windows 11's native Sound Recorder utility provides basic local audio capture with integrated speech-to-text.
- Free Quota & Limits: Free system utility with unlimited local microphone and system audio recording. A 3-hour limit applies per continuous recording file, but total monthly recording has no cap. Basic speech-to-text generation operates locally.
- Speaker Identification: Not supported; generates single-channel audio transcription.
- In-Person Support: Fully supported for audio captured via laptop or USB microphones.
- Export Restrictions: No export walls; output is saved locally as audio files and text copy.
Where it caps out: Windows Sound Recorder handles basic audio capture competently, but the speech-to-text output is single-stream with no speaker separation. For a solo practitioner recording voice memos or dictation, this is functional. For a team meeting with five participants around a conference table, the resulting transcript is an unlabeled monologue that requires extensive manual reconstruction.
Category 3: Open-Source Local AI Utilities
For users with sensitive audio who refuse to upload conversations to cloud servers, open-source Whisper-based desktop applications provide offline transcription with zero recurring costs—at the expense of setup complexity and feature access.
8. Local Whisper Desktop GUIs (MacWhisper Free & Buzz)
Both MacWhisper Free and Buzz wrap OpenAI's open-source Whisper speech recognition model into desktop applications that run entirely on local hardware.
- Free Quota & Limits: 100% free with unlimited local transcription processing on PC or Mac hardware. There are no monthly caps; speed depends on the user's GPU or CPU.
- Speaker Identification: Mixed. MacWhisper Free locks speaker diarization behind its €59 Pro lifetime tier. Buzz provides basic offline transcription but requires manual speaker tagging after processing.
- In-Person Support: Complete offline capability; processes pre-recorded audio files (MP3, WAV, M4A) without any internet connectivity.
- Export Restrictions: Buzz exports SRT and VTT subtitle files for free. MacWhisper Free limits advanced subtitle formatting and batch processing unless upgraded to Pro.
Where it caps out: Setting up a local Whisper transcription pipeline requires downloading model files (ranging from hundreds of megabytes to several gigabytes), accepting slower processing speeds on CPU-only machines, and living without automated speaker identification unless paying the upgrade fee. For a privacy-conscious journalist transcribing a sensitive interview recording, the trade-off is worth it. For a sales rep who needs a transcript ready five minutes after a call ends, the manual workflow creates friction that cloud tools eliminate.
What "Free" Usually Costs You: Three Operational Bottlenecks
Each free tool described above solves a real problem. None of them are scams or dead ends. But every user who has relied on free AI note takers for more than a few weeks encounters at least one of these three structural limitations—usually at the worst possible moment.
Bot Dependency and Meeting Entrance Friction
Cloud meeting bots (Otter, Fireflies, tl;dv) must explicitly join video calls as participants. This creates two friction points.
Top 5 BEST Bot-Free AI Meeting Note Taking Apps
First, meeting hosts increasingly reject third-party bot admission requests out of security policy or privacy discomfort. Platform-level changes compound this: Google Meet's guest admit queues, for example, hold unauthenticated bots in a waiting room until a host manually approves entry. In visual demonstrations of this friction, animated interfaces show bots getting stuck in waiting rooms with the status STUCK IN THE WAITING ROOM, requiring manual intervention, or remaining STILL IN THE CALL long after human participants have disconnected.
Second, unannounced recording bots entering sensitive client calls or internal strategy meetings create social discomfort that no transcription quality justifies. A bot cannot explain its data handling practices to a skeptical client.
Channel Blindspots: In-Person Meetings and Phone Calls
Software meeting bots are architecture-bound to web conferencing platforms. They cannot capture offline environments: a coffee shop consultation, a hallway negotiation, a physical boardroom discussion with no video bridge.
Cellular and landline phone calls present the same blindspot. Desktop web bots lack the hardware pathway to tap into direct phone call audio without complex third-party VoIP routing configurations. A recruiter conducting phone screens or a journalist recording a source interview over a cellular call finds that every cloud bot in Category 1 is structurally incapable of capturing that conversation.
Feature Striping and Hard Execution Stops
Free tiers do not simply offer "less"—they strategically remove features that make transcripts actionable.
Speaker diarization disappears on tl;dv's free plan, Apple Voice Memos, and MacWhisper Free. The user receives a transcript but must manually assign every line to a speaker, turning a 30-minute meeting into an hour of forensic editing.
Export walls block downloads on Fireflies' free tier entirely. The transcript exists inside the web app but cannot leave it, forcing users into copy-paste workflows that defeat the purpose of automation.
Mid-meeting cutoffs are the most damaging failure mode. When Otter Basic hits its 30-minute per-conversation limit, transcription stops without warning. The call continues normally for participants, but every decision, commitment, and action item spoken after minute 30 is lost. No recovery is possible without upgrading.
For teams evaluating more structured solutions to these workflow constraints, the AI Voice Recorders & Apps Guide provides a deeper analysis of how different capture architectures handle multi-speaker environments.
Choosing the Right Capture Architecture: Comparative Matrix
To determine whether a free software tool fits your specific meeting volume, environment, and security requirements, evaluate the structural differences across categories. Free AI meeting bots and phone apps solve different problems than dedicated hardware recorders, and the right choice depends entirely on where and how you record.
| Dimension | Free AI Meeting Bots | Free Phone OS Apps | Dedicated Hardware Recorder |
|---|---|---|---|
| Recording Method | Calendar bot joins Zoom/Teams/Meet calls | On-device phone mic capture | MagSafe magnetic hardware attachment with dual acoustic sensors (e.g., UMEVO Note Plus) |
| In-Person Meetings | No (virtual web meetings only) | Yes (near-field microphone) | Yes (far-field acoustic sensors for room audio) |
| Phone Calls | No | Limited (carrier and OS call recording restrictions apply) | Yes (dual-engine sensor captures physical phone speaker and mic vibrations) |
| Monthly Minute Limit | 300 min/mo or storage capped | Unlimited local storage, dependent on phone capacity | 1-year MAX subscription included, subject to fair-use terms; 64GB local storage supporting up to 40h continuous recording |
| Speaker Labeling | Varies (often locked on free tiers) | Pixel: Yes (US English); Apple: No | Yes (140+ languages with automated multi-speaker identification) |
| Privacy & Compliance | Cloud server processing; bot visible to all attendees | On-device processing | GDPR / ISO 27001 / SOC2 / HIPAA compliant architecture |
For users evaluating hardware options in this category, the Plaud Note Alternatives Guide compares dedicated recorders across comparable price points. For those assessing whether a hardware upgrade is warranted, Why Upgrade Your Voice Recorder in 2026 provides architectural context. Specialized recording workflows for meetings and interviews benefit from purpose-built capture hardware.
Summary
Zero-budget AI note takers provide practical solutions for light meeting schedules. If you conduct fewer than five short virtual meetings per month, cloud bots like Otter or Fireflies cover basic transcription needs without immediate cost. For offline notes or personal dictation on supported hardware, native utilities like Google Recorder or local Whisper applications offer uncapped transcription without subscription fees or cloud exposure.
Each free option caps out at a specific point—a 30-minute conversation limit, a 400-minute storage ceiling, a 10-summary lifetime allocation, a missing speaker label feature. The key to avoiding lost meeting data is knowing which cap applies to your specific recording environment and meeting volume before you hit it.
Review your monthly meeting logs. If you frequently approach 30-minute call durations, need reliable recording for offline interviews and phone calls, or work in environments where meeting hosts routinely block third-party bots, evaluate dedicated hardware capture alongside your software tools. The goal is not to spend more—it is to ensure that when a meeting matters, the last ten minutes of decisions do not vanish silently.
FAQ
Is there a completely free AI note taker with no monthly minute limits?
Yes. Native phone utilities like Google Recorder on Pixel devices and Apple Voice Memos on iOS 18 and newer provide completely free speech-to-text transcription without monthly minute caps. Open-source desktop utilities including Buzz and MacWhisper Free also process pre-recorded audio files locally on your computer with no time restrictions. These tools achieve unlimited usage by running transcription models on-device rather than routing audio to paid cloud infrastructure.
Can free AI note takers record in-person meetings?
Mobile operating system apps including Google Recorder and Apple Voice Memos can capture in-person conversations using your device's built-in microphone. Desktop recording software like Windows Sound Recorder performs the same function through laptop or USB microphones. However, web-based AI meeting bots such as Fireflies or tl;dv are designed exclusively for virtual calls on platforms like Zoom, Teams, and Google Meet, and cannot capture physical offline meetings where no video bridge exists.
Do free plans limit monthly minutes?
Most cloud-based AI meeting assistants enforce strict limits on free tiers. Otter.ai's Basic plan caps transcription at 300 total minutes per month and limits individual conversations to 30 minutes. Fireflies.ai provides unlimited transcription minutes on its free tier but limits total team storage to 400 minutes and blocks transcript downloads. tl;dv allows unlimited recording but caps AI summary generation at 10 lifetime uses. Native phone apps and open-source local tools generally do not impose monthly minute limits because processing happens on-device at no incremental cost to the provider.
What happens when a free AI note taker exceeds its minute cap during a meeting?
When a session hits a duration cap, such as Otter's 30-minute conversation limit on the free plan, the AI bot stops transcribing immediately. The call continues normally for human participants, but audio spoken after the limit is reached will not be recorded, transcribed, or included in automated AI summaries. The cutoff typically provides no audible warning to meeting participants, meaning the user discovers the gap only after the meeting ends and the incomplete transcript appears. The only recovery path is to upgrade the account tier before the next call begins.
References
- Pricing & Plans | Otter.ai — Otter.ai
- Pricing & Plans | Fireflies.ai — Fireflies.ai
- What are the limits of the Free plan? | tl;dv Help Center — tl;dv
- What is Notion AI? FAQs | Notion Help Center — Notion Labs, Inc.
- Record and transcribe audio in Notes on iPhone | Apple Support — Apple Inc.
- Use speaker labels in a recording (Pixel 6 & later) | Google Pixel Help — Google LLC
0 comments