Turn recordings into transcripts and summaries in minutes
Upload audio or video for multilingual transcription, AI notes, and action items
5 Cantonese Speech-to-Text Tools Tested for 2026: Which Free Tier Actually Works?
First, Get Clear: Do You Want to 'Understand a Recording' or 'Find That Meeting Later'?
I have about 15 meetings a week, and two or three of them are with Hong Kong colleagues and clients—the whole meeting is mostly in Cantonese, with English mixed into the sentences. My old approach was to record everything and listen again afterward. The problem: a two-hour meeting means another two hours of listening and organizing. By the time I finish the meeting notes, a whole day has passed.
If you're also looking for a free Cantonese speech-to-text tool, here's my conclusion upfront: free plans aren't unusable, but you first need to figure out which pain point you're stuck on.
The first is 'understanding in the moment.' Sentences are too long, the accent is unfamiliar, or it's mixed Chinese-English, and you need to see text immediately.
The second is 'finding it later.' Two months later your boss asks, 'What did the Hong Kong client say about the delivery date last time?' and you need to dig that one sentence out of a three-hour recording.
The third is 'handing it off.' Those action items after the meeting—who's responsible, when it's due—are gone as soon as they're said out loud.
These three pain points actually require different capabilities: the first needs real-time transcription, the second needs search and Q&A, and the third needs automatic action item extraction. And most Cantonese transcription tools that advertise 'free' only solve the first one.
Before Choosing a Free Cantonese Transcription Tool, Understand These 4 Key Points
1. Is the free unit 'minutes' or 'files'? Some tools say free, but the limit is only the first 15 minutes per file; others give you a fixed number of minutes per month, and once it's used up, it's gone. You need to calculate roughly how much you'll record in a month to know whether that allowance is useful or just looks good on paper.
2. Does it actually use a Cantonese model, or force a Mandarin model onto it? This is the biggest pitfall in Cantonese recognition. A Mandarin model encountering spoken Cantonese often outputs sentences that don't make sense or use strange characters, especially when a sentence contains mixed Chinese-English phrasing like 'check 一下個 file,' which often comes out as Chinese characters with similar pronunciation. The way to judge is simple: try it with a real recording of your own, not the official demo audio.
3. Does it only give you a transcript, or something you can use after the meeting? A transcript is raw material, not a finished product. Whether it can automatically produce summaries, chapters, and action items, and whether you can ask in one sentence 'What was the conclusion of this meeting?'—the difference is whether you spend 5 minutes or 50 minutes afterward. This is the area free tools most often cut.
4. Cross-platform and team collaboration You record on your phone today and organize on your computer tomorrow. If the tool only supports a single platform, just transferring files is annoying enough. If you also need to share with your team, free personal plans are almost never enough—more on that later.
Tinrec—The One I Settled On After Testing
Tinrec is an AI meeting notes and collaboration tool for individuals and teams, covering iOS, Android, and web, while the desktop version can directly handle online meetings. I kept it mainly not because it's free, but because it solves the second and third pain points I mentioned above together.
During the meeting: no need to invite a bot. We use Google Meet for meetings with the Hong Kong team and Teams with clients. Tinrec's desktop version directly captures the computer's system audio, so no extra meeting bot needs to join as a participant. This is very practical for internal meetings—you don't want a strange account showing up on the list in front of a client. While I'm in the meeting, the transcript appears in real time, so I don't panic if I'm called on unexpectedly.
After the meeting: what it gives you isn't a transcript, it's a usable package of materials. Summaries, chapters, key points, and action items are generated automatically; commonly used team terminology can be added to a hotword list first so proper nouns aren't turned into strange characters every time. When you need bilingual comparison, it can display the original text, translation, or bilingual version during recording. These things used to require manual work; now it's basically opening the file, scanning it, fixing a few words, and sending it out.
The biggest time-saver: just ask it. This is where I think it differs most from most free tools. After recording, you can ask questions directly about the meeting content, such as 'What date was finally agreed for delivery?' or 'Who's responsible for this quote?' It answers based on semantics, not by dumping a bunch of keyword search results on you. Last month my boss suddenly asked me in a meeting about a decision on a project from last quarter. I opened it and asked on the spot, and it took less than a minute—previously in moments like that I'd probably just say, 'Let me check and get back to you.'
Stop organizing recordings by hand
Upload audio or video and automatically get a transcript, summary, and action items
Test conditions, stated clearly. I'm not going to report a pretty accuracy number here, because it depends entirely on your recording quality. My test method was: take the same 8-minute real Cantonese meeting recording (phone audio, air conditioning and keyboard noise in the meeting room, mixed Chinese-English), run it through several tools, and compare. Under those conditions, Tinrec's advantage isn't zero errors, but that it captures the structure of mixed Chinese-English sentences more completely, and after adding proper nouns to the hotword list it's much more stable. Actual results will still be affected by accent, multiple people speaking at once, and recording distance.
I'll mention the downsides too. The free version is a basic allowance—fine for testing the waters, but not enough as a daily workflow; heavy users will need to pay. Also, its strength is the meeting itself, not batch transcription of large numbers of files; if you have dozens of hours of historical recordings to process at once, that type of tool will suit you better.
About the team version. If you want to use it for an entire department, Tinrec's team version is a separate team space, not 'everyone sharing one personal account.' Meeting data belongs to the team, and when a member leaves or loses their seat, the data isn't taken away by the individual and needs to be handed over to other members. Roles and seats should be viewed separately: members without a seat can still view, play, and read team data, but cannot upload, edit, or export. The admin side has member management, usage analytics, and audit logs. Pricing is currently USD 29.80 per paid seat per month, or USD 199 per paid seat per year, which works out to about USD 16.58 per month.
Who it's for. If you frequently handle Chinese or mixed Chinese-English meetings, need to produce minutes and action items directly after the meeting, and also want to dig out details through Q&A after the meeting ends, Tinrec is the smoothest set I've used so far.
Besides Tinrec, What Other Options Are There?
Subanana. A Cantonese transcription tool made by a Hong Kong team, focused on preserving the original English when Chinese and English are mixed. The first 15 minutes of each file can be previewed for free, and the output formats are fairly complete. Its positioning leans toward 'turning a recording into a transcript'; summaries, action items, cross-meeting Q&A, and team data retention after transcription aren't its focus. If you only occasionally need a Cantonese transcript, it's enough; but if what you want is meeting data you can keep asking questions about, Tinrec's AI Q&A and team space are things it doesn't have.
Google AI Studio (Gemini). This is a genuinely free route with decent Cantonese performance, and you don't have to deal with complex API setup. But at its core it's a conversational AI: you have to prepare the recording yourself, give it instructions to transcribe, then give a second set of instructions to organize it. It has no meeting database, no action item list, no team permissions—every time starts from zero. It's a great deal for people who don't mind manual operation; for people who have meetings every day, just repeatedly pasting instructions eats up the time saved.
Otter.ai. A mature choice for English meetings, with a fairly generous free tier, but its long-standing strength is English business meetings, and Cantonese support is relatively weak. If 90% of your meetings are in English, it will be more cost-effective than Tinrec—this is one of the few scenarios where I'd recommend another tool; but as soon as a lot of Cantonese appears in the meeting, you'll run into the same reason we switched tools back then.
There are also built-in phone dictation and Google Docs voice typing, which are completely free and have zero learning curve, but they can only input in real time and can't import existing recording files, so they're not usable for formal meetings—better for memos and short dictation.
Pitfall Guide: The 4 Most Common Mistakes When Choosing Free Cantonese Transcription
Pitfall 1: Only looking at the word 'free' and not the free unit. Some are the first 15 minutes per file, some are a fixed number of minutes per month—the difference is huge. Calculate using your own recording length, not the official demo file.
Pitfall 2: Using a Mandarin model for Cantonese. This is the most common reason for failure. The test isn't reading ad copy; it's running a conversation you'd actually have (ideally with mixed Chinese-English) and seeing whether English words get written as Chinese characters with similar pronunciation.
Pitfall 3: Taking only the transcript and leaving. If you only use the tool as speech-to-text, you're using only a small part of its capability. What really saves time is the later part: summaries, action items, Q&A.
Pitfall 4: Forgetting to ask for consent. Recording and transcription involve personal data, especially when meetings are cross-regional. It's best to explain at the start that the meeting will be recorded and what it's for, and get participants' consent when necessary.
Conclusion: Which One Should You Actually Choose?
- Need to handle Cantonese and mixed Chinese-English meetings, plus post-meeting summaries and action items → Tinrec
- Want to find details directly through Q&A after recording → Tinrec (this is where it differs most from most free tools)
- Need to use it on both phone and computer, and import existing audio files → Tinrec
- Need to share with a team, with member and seat management → Tinrec team version
- Meetings are almost entirely in English and you just want to keep costs down → Otter.ai is worth considering
- Only need a pure transcript with no further processing → a file-based tool like Subanana is enough
- Don't want to spend anything and are willing to give instructions manually → Google AI Studio is a viable route, but you have to accept it has no database or workflow
My advice is not to rush into paying. Use Tinrec's free version and run it on the Cantonese recording that hurts the most—meeting notes are the kind of thing where one try tells you whether it can save your evening.
References
- Free Cantonese Speech-to-Text App 2026 | Subanana
- Speech to Text - Google Play App
- SoundType AI Ultra-Precise Transcription, Speech to Text, AI Dictation - Google Play App
- 2026 Latest 5 Cantonese Speech Recognition Transcription Software Recommendations: Solving Cantonese Meeting Notes and Mixed-Language Pain Points - Tinrec Blog
- 'Free + Cantonese OK: How to Use Google AI Studio to Turn Your Recording into Text' Finding a good transcript app is painful; I don't know if you've tried different recording-to-text apps—either fake free, charging you a big price after a few dozen minutes, or only supporting English and not Cantonese, or requiring a whole API setup before you can use it // Recently I found that Google Gemini is world-class in this area: - [Truly free] No subscription fee after a few uses - [Cantonese-friendly] Gemini is accurate
Turn every recording into actionable outcomes
Get 60 free transcription minutes when you sign in. No credit card required.
Related Reading
You might also like

How to Choose Cantonese Speech-to-Text in 2026: 5-Step Test of Free Tools
Many Cantonese speech-to-text tools claim to be free, but file limits, language support, and post-meeting organization vary widely. This article uses 5 key steps to test and compare, and explains who Tinrec, Subanana, cSubtitle, and Google AI Studio are best for.

2026 Guide: Free Cantonese Audio to Text Solutions — Tinrec Free Tier + AI Meeting Summaries
This guide helps you avoid common pitfalls when converting Cantonese audio to text for free. We test Tinrec's free tier, transcripts, AI summaries, and team collaboration, and compare alternatives like Notta, Otter, and Google AI Studio to help you find the best Cantonese transcription workflow.

Tinrec vs Notta 2026: 5-Dimension Comparison for Cantonese Meeting Transcription
In 2026, choosing a Cantonese audio-to-text tool isn't about which is most accurate, but which workflow fits you best. This article compares Tinrec and Notta across 5 dimensions, covering differences like bot-free meeting recording, AI Q&A, and team spaces, plus use cases for Otter.ai and TurboScribe and 3 common pitfalls to avoid.

5 Cantonese Audio-to-Text Tools Tested and Compared in 2026: How to Solve the Transcription Pain Points Dcard Users Ask About Most
Discussions on Dcard about converting Cantonese audio to text mostly focus on two pain points: tools that don't understand Cantonese and privacy concerns about uploading files. This article starts from real-world use cases, compares the differences among 5 common approaches, and uses Tinrec as an example to explain how real-time transcription, bot-free meeting recording, AI summaries and Q&A, and team spaces solve these problems. It ends with 5 buying checkpoints and FAQs.

4 Cantonese Speech-to-Text Tools Compared for 2026: Free vs Paid Options Explained
How to choose a Cantonese speech-to-text tool in 2026? This article addresses common misconceptions and compares Tinrec, Otter.ai, PLAUD, and Notta on Cantonese support, free tiers, and post-meeting organization. It also provides 4 key buying factors and 3 pitfalls to avoid, helping you find the best transcription solution.

5 Best Cantonese Speech-to-Text Tools in 2026: Which Free Tier Actually Works?
Starting from a university group project with Cantonese discussions, this hands-on comparison tests free Cantonese speech-to-text tools on free quotas, mixed Chinese-English recognition, and post-meeting organization, sharing real-world experience using Tinrec in campus classes, interviews, and group collaboration.

Which Cantonese Audio-to-Text Tool Is Best? 2026 Hands-On Test of 5 Options, Top Pick for Chinese Meetings
The biggest pitfall when choosing a Cantonese audio-to-text tool is relying only on the official language list. This article first clarifies common misconceptions, then provides a hands-on testing process you can run yourself, and uses Tinrec to demonstrate a complete workflow from recording and transcription to summaries, action items, and team archiving, including pricing and plan limitations.

How to Choose Cantonese Audio-to-Text in 2026: 5 Steps to Pick the Right Free Tool
How do you choose a Cantonese audio-to-text tool? This article reviews 6 free and paid options in 2026, including Tinrec, Subanana, cSubtitle, Google AI Studio, Notta, and TurboScribe. It walks you through 5 steps—from language support and live meetings to post-meeting organization and team sharing—to find the one that saves you the most time.

2025 Top 5 Cantonese Speech-to-Text Tools Tested: Which One Turns Transcripts into Usable Data?
After hands-on testing of several Cantonese speech-to-text tools in 2025, the key takeaway: getting a transcript is just the beginning. This article breaks down the complete workflow for Cantonese audio transcription, from real-time transcription and handling mixed Chinese-English to AI summaries and action items, team sharing, and purchasing considerations, with Tinrec demonstrating each step.
