4 Cantonese Speech-to-Text Tools Tested: Which Free Version Actually Works?

When you type "free Cantonese speech to text" into the search bar, what you're really asking is: Which tool's free version can actually be used for work, rather than hitting a paywall after a few uses?

There's no standard answer because "free" means different things for each tool. Some offer a free trial for the first 15 minutes, some let you transcribe for free but not export, and others are free with no limits but very basic features.

This article first breaks down the differences using four key criteria, then compares the four most frequently mentioned tools: Tinrec, Subanana, cSubtitle, and Atter App.

Why is it so hard to choose a free Cantonese speech-to-text tool? — You don't need a tool, you need a solution

Bottom line: most people get stuck not because they can't find a tool, but because the tool they find only solves a third of the problem.

The first scenario is meetings. You have a two-hour Cantonese meeting recording, but you have to transcribe it yourself, and then organize who's responsible for what. Your time isn't spent in the meeting; it's spent after the meeting.

The second scenario is interviews or lectures. Cantonese content often includes English proper nouns and brand names. General tools often make mistakes in mixed Chinese-English segments, and correcting errors is more tiring than retyping.

The third scenario is data that scatters after use. The recording sits on someone's phone, the transcript is on someone's cloud drive, and next time you need to check "that budget number the client mentioned last time," no one can find it.

If you've experienced any of these, what you really need isn't just "turning sound into text," but making that content searchable, queryable, and transferable later.

Before choosing a free Cantonese transcription tool, understand these 4 key points

1. What kind of "free" is it?

There are three common types: free preview (e.g., first 15 minutes of each file), free transcription but paid export, and free with limited features.

Before downloading, confirm whether you need to "see it" or "take it away." If you can't export the transcript, its help for your workflow is actually very limited.

2. Cantonese recognition isn't just about understanding; it also needs to distinguish speakers

The real challenges of Cantonese transcription lie in details: spoken vs. written output, traditional vs. simplified characters, mixed Chinese-English proper nouns, and automatic segmentation when multiple people talk over each other.

These differences are usually invisible, but they all surface in real meetings.

3. Is there a next step after transcription?

A tool that only gives you a transcript is like handing you 30 pages of raw material.

Tools with summaries, chapters, action items, and AI Q&A turn that material into usable meeting notes. The gap here often affects your daily time more than recognition accuracy.

4. Will you switch devices or need to share with others?

If you record on iPhone today and organize on computer tomorrow, and the tool doesn't support cross-platform, your data will be stuck on one device.

For team use, check whether it allows meeting data to belong to the team, rather than leaving with a colleague's account.

Tinrec — The top choice for turning Cantonese recordings into usable data

Tinrec is an AI meeting notes and collaboration tool for individuals and teams. It supports live recording, online meeting transcription, audio/video file import, and real-time translation, and can automatically generate summaries, chapters, action items, and mind maps.

It's not just speech-to-text; it turns meetings into searchable, collaborative, actionable team data.

During the meeting: No extra bot needed

Tinrec desktop version directly captures your computer's system audio, handling Zoom, Google Meet, Microsoft Teams, Webex, and other online meetings without an extra meeting bot joining.

In practice: open Tinrec before the meeting starts, and the transcript grows as the meeting progresses. You don't have to listen and take notes simultaneously, and you won't discover afterward that the key part wasn't recorded.

After the meeting, different speakers are labeled, and you can rename them, making it clear who said what.

Post-meeting organization: Summaries, chapters, and action items all at once

As soon as the meeting ends, AI summaries and chapters are generated. You can see which topics were discussed and where conclusions were made, and jump directly to them.

Action items are also extracted from the conversation—who's responsible and when it's due—much faster than going back through the recording yourself.

Even better is AI Q&A. You can directly ask "Who mentioned the budget in the last meeting?" or "What did the client say about the delivery date?" It answers based on content semantics, not just a list of keyword search results.

Content processing: Translation, export, and regeneration

If not all participants in a Cantonese meeting are Cantonese speakers, real-time translation lets you switch between original, translated, or bilingual views.

Output supports transcripts, subtitle files, and document formats, and can be exported to Notion, Google Docs, OneNote, Dropbox, etc. for further use. When needed, you can also generate reports, tables, and meeting minutes documents.

About accuracy, I want to be honest

Tinrec does not promise a fixed accuracy rate because recognition is affected by accent, speech rate, background noise, overlapping speakers, and technical terms—especially for Cantonese.

My advice: don't trust any official numbers. Use your own Cantonese recording to test the free quota—that's the most realistic result for your scenario.

Three reasons I recommend it

1. No bot recording: No extra account as a participant in online meetings, and no need to coordinate with the host in advance.
2. Complete post-meeting workflow: From transcript, summary, action items to AI Q&A and multi-format export, all done in one place without moving files between tools.
3. Team version keeps data: Meetings belong to the team, with member roles, seats, usage analytics, audit logs, and audio recycle bin. When members leave, data can be transferred and won't disappear with personal accounts.

Limitations to note

The free version provides basic quota, suitable for trial and light use. If you're a heavy user with many meetings per week, you'll need to consider weekly pass, Pro, or team version.

Also, the team version is billed by "seat." Members need a seat to record, upload, and edit; members without a seat can still view, play, and read team data.

Who is it for

If you need to handle Chinese (including Cantonese) and mixed Chinese-English meeting content, don't want a bot joining online meetings, want to use AI to find key points directly after recording, or your team shares meeting data, Tinrec is currently one of the most suitable choices.

Besides Tinrec, what other free Cantonese transcription options are there?

Subanana: A Cantonese transcription service developed by a Hong Kong team, supporting mixed Chinese-English, spoken or written output, and traditional/simplified character switching.

Free accounts can upload files and preview the first 15 minutes of transcription per file, but exporting files is a paid feature. Note that it focuses on personal file transcription, without desktop bot-free meeting recording, team spaces, seat allocation, or usage auditing.

cSubtitle: No account registration required; upload files and start transcribing. Supports batch processing, with a single file limit of 4GB and 5 hours, focusing on Chinese processing.

Its strength is processing many files at once. But it's more of a transcription tool than a meeting tool—no live recording transcription, no AI Q&A or action item extraction, and it can't consolidate content into a shared team database.

Atter App: Focuses on being free, ad-free, and without quota limits. Can import from phone recordings, photo album videos, or Apple Watch. Provides transcripts, summaries, chapters, and AI Q&A, friendly for those frequently handling interviews and lecture notes.

However, the original author's testing mentions recognition accuracy of about 70%, with common brand and person names prone to errors. It leans toward mobile personal recording; Tinrec focuses on online meetings and team data consolidation, and can maintain professional terminology recognition with team hotwords.

Pitfall guide: 3 most common traps when choosing a free Cantonese transcription tool

Pitfall 1: Mistaking "free trial" for "free use"

Many tools' free versions are "preview first 15 minutes" or "can transcribe but not export." To actually take the file, you have to pay. Check the free scope—minutes, features, or export rights—before investing time.

Pitfall 2: Only looking at official accuracy numbers

Those numbers are measured in a quiet recording studio, which is different from your meeting room—with air conditioning, keyboard sounds, and people talking over each other, the numbers will drop. Testing with your own recording is more accurate than any marketing.

Pitfall 3: Treating the tool as just a typewriter

If you only use it to "turn sound into text," you're using only a third of its capabilities. After recording, directly ask AI to summarize, extract action items, and answer questions—saving much more time than transcription itself.

Final reminder: Before recording, confirm compliance with local laws and obtain consent from participants when necessary. This is more important than which tool you choose.

Conclusion: Which one should you choose?

• Just occasionally transcribe a Cantonese recording, zero budget → Subanana (free preview), cSubtitle, Atter App are all worth trying.
• Need to record online meetings without an extra bot participant → Tinrec.
• Want to ask AI directly "What's the conclusion?" "Who's responsible for what?" after recording → Tinrec (has AI Q&A and action item extraction).
• Need to continue using on phone, computer, and web → Tinrec.
• Team needs to share meeting data and manage members, seats, and usage → Tinrec team version.
• Need to process many long files at once → Batch transcription tools like cSubtitle are more suitable.

Simply put, if you just want to "free turn Cantonese into text," there are many options; but if you want "usable data right after the meeting," Tinrec is the most complete among these.

I suggest opening Tinrec's free version first, testing with your own Cantonese meeting recording, and considering an upgrade only if it works for you. No need to pay upfront.

References

• Free Cantonese speech-to-text app 2026 | Subanana
• Online AI speech, video, audio file auto transcription - Free trial - cSubtitle
• AI speech to text fast audio file online transcription, transcripts and video subtitles - Free trial - cSubtitle
• "Free + Cantonese OK: How to use Google AI Studio to transcribe your recordings" Finding a good transcript app is painful; I don't know if you've tried different recording-to-text apps—either fake free, charging a lot after a few minutes, or only supporting English, not Cantonese, or requiring a lot of API setup. Recently I found Google Gemini is world-class in this area: - [Truly free] No subscription after a few uses - [Cantonese friendly] Gemini accurate
• Free unlimited AI speech-to-text tool, say goodbye to transcript organization troubles: Atter App | Lin Weiqing's Salon