Turn recordings into transcripts and summaries in minutes
Upload audio or video for multilingual transcription, AI notes, and action items
4 Cantonese Speech-to-Text AI Tools Compared (2026): How to Choose for Cantonese Recognition and Mixed Chinese-English
It's Wednesday afternoon, and your manager forwards a 47-minute client interview recording. It's entirely in Cantonese, with English product names and a few internal code names mixed in. You need to deliver a transcript and key points by tomorrow morning.
In the past, this meant three hours of re-listening with headphones. Now everyone knows AI can handle it, but the question becomes: which tool actually understands Cantonese?
This article compiles currently available public information into a single table, clearly laying out free quotas, export formats, and limitations, so you can decide which one is worth your time to upload your own audio and test.
Three Stages That Affect Cantonese Transcription
Turning a recording into a deliverable document involves three stages: upload, recognition, and organization. Each has its own hurdles.
Upload determines what you can throw at it. Web-only upload versus support for computer system audio, YouTube links, or mobile recordings—the applicable scenarios are completely different.
Recognition is the most condition-sensitive stage. Cantonese has many colloquial characters (唔, 係, 嗰, 嘅), and Hong Kong workplaces commonly mix Chinese and English. Forcing a single-language model often produces garbled text or inaccurate results.
Organization determines whether the output can be used directly. A bare transcript versus one with paragraphs, punctuation, speaker separation, summaries, or even action items—the usage cost differs greatly.
Note: the above are common factors affecting results, not speculation about any tool's technical implementation.
The 4 Tools Compared
Four tools are included, with clear positioning differences:
- Tinrec: An AI meeting notes and team collaboration tool, emphasizing bot-free recording of online meetings, post-meeting summaries, and team knowledge retention. This article cites information provided by the product team; this disclosure is made here.
- Subanana: Focused on Cantonese transcription, supports file upload, YouTube link pasting, and browser recording.
- CantoSub: Highlights multiple Cantonese branches and personalized dictionaries.
- Notta: Multilingual transcription and meeting notes, with a well-structured plan lineup.
Comparison Methodology and Conditions
- Data verification date: August 2026, based on official product pages and public information.
- No standardized timing experiments were conducted, so no measured numbers such as upload seconds or transcription seconds are provided; this is a data comparison, not a benchmark article.
- Free quotas, prices, and features may change with plan adjustments; please refer to the official website's current page for actual conditions.
- Items that cannot be confirmed from public information are marked as "pending confirmation" and not filled in with speculation.
Feature and Quota Comparison Table
| Item | Tinrec | Subanana | CantoSub | Notta |
|---|---|---|---|---|
| Cantonese support statement | Covers Chinese and multilingual meetings, can organize mixed Chinese-English content | Cantonese model, handles mixed Chinese-English | Hong Kong Cantonese, Guangzhou dialect, Dongguan dialect, Shunde dialect, etc. | Multilingual transcription (actual Cantonese performance to be verified) |
| Input sources | Live recording, desktop system audio capture, audio and video files, mobile, some web links | Upload audio or video, YouTube links, browser recording | Speech to text (per official website) | File upload and meeting scenarios |
| Output formats | Transcript, summary, chapters, action items, AI Q&A, translation, multiple formats and third-party exports | Speaker-separated, punctuated text, optional colloquial or written style, traditional or simplified output | Colloquial to written conversion, traditional Chinese output | Transcript, translation, summary |
| Export formats | Multiple formats and third-party tool exports | SRT, VTT, TXT, DOCX, XLSX, Markdown | .txt, .docx, .pdf | Varies by plan |
| Free quota | Personal free version (quota per official website); team first trial 7 days, 1 free seat, 300 minutes shared import | First 15 minutes per file free preview, export is a paid feature | New users 30-minute trial, no credit card required | 120 minutes per month, max 3 minutes per single item |
| Paid starting price | Team USD 29.80 per paid seat per month; annual USD 199 (approx. USD 16.58/month) | Pending confirmation | Pending confirmation | Pro annual equivalent approx. USD 8.17/month |
Observations and Use Cases for Each Tool
Tinrec: End-to-End Meeting Workflow, Cantonese Verified with Your Own Audio
Tinrec is not just a transcriber; it turns meeting recordings into searchable, queryable, exportable data. The desktop version directly captures computer system audio, so when handling online meetings like Zoom, Google Meet, Teams, or Webex, no extra bot needs to join as a participant.
The post-meeting stage is its differentiator: beyond transcripts, there are summaries, chapters, action items, mind maps, AI Q&A, real-time translation, and export channels to tools like Notion and Google Docs. The team version stores data in a team space, with admin controls for roles, seats, usage, and audit logs.
Note: The product materials emphasize Chinese and multilingual meetings and mixed Chinese-English content organization, but do not promise specific Cantonese accuracy. Cantonese accents, proper nouns, and multiple people talking over each other will affect results; it's recommended to test a segment with your own recording first.
Stop organizing recordings by hand
Upload audio or video and automatically get a transcript, summary, and action items
Who it's for: Those who need Chinese or Cantonese meeting notes, want post-meeting summaries and action items, and need meeting data retained within the team.
Subanana: Especially Smooth When You Have a YouTube Link
Subanana's workflow is short: upload an audio or video file, paste a public link, and AI outputs a speaker-separated, punctuated transcript. Mixed Chinese-English is listed as supported, and you can choose colloquial or written style, traditional or simplified.
Free accounts can upload and preview the first 15 minutes of each file without a credit card; exporting files is a paid feature. The official site also notes that accuracy decreases in noisy environments or when multiple people talk over each other.
Who it's for: Those who frequently need to quickly turn public videos or short interviews into readable transcripts. For long meetings, be mindful of the 15-minute preview limit.
CantoSub: Cantonese Branches and Dictionary Settings Are Key
CantoSub explicitly lists support for Hong Kong Cantonese, Guangzhou dialect, Dongguan dialect, Shunde dialect, and other branches, and offers personalized dictionary settings. It converts colloquial Cantonese to written Chinese, outputs traditional Chinese, and can export .txt, .docx, .pdf. New users get a 30-minute free trial with no credit card required.
If your recording is not a Hong Kong accent but rather the speech style of Guangzhou, Dongguan, or Shunde, this "branch" labeling is more valuable for reference when choosing.
Who it's for: Those who need to handle specific Cantonese branches or want to use custom dictionaries to correct technical terms.
Notta: Clear Plan Tiers, Cantonese Needs Your Own Testing
Notta covers multi-platform transcription, meeting notes, file import, translation, and team plans. The free version offers 120 minutes per month with a maximum of 3 minutes per single item. Pro annual billing is approximately USD 8.17/month for 1,800 minutes per month. Business annual billing is approximately USD 16.67 per seat per month and includes role permissions and usage reports.
Its plan structure is budget-friendly, but public information does not provide independent Cantonese-specific details; actual recognition performance needs to be verified with your own audio files.
Who it's for: Those who need multilingual transcription and prefer plans with clear minute allowances and permission tiers.
Price and Available Quota Comparison
The following is public information verified in August 2026. Please check the official website for current conditions before paying.
| Tool | Plan | Billing cycle | Quota or benefits | Notes |
|---|---|---|---|---|
| Tinrec | Free | Personal | Basic quota to experience recording, import, summary, and transcription | Quota per official website |
| Tinrec | Team trial | 7 days | 1 free seat, 300 minutes team shared import | First eligible team |
| Tinrec | Team monthly | Per paid seat/month | USD 29.80 | 2,000 minutes shared import per seat per month |
| Tinrec | Team annual | Per paid seat/year | USD 199 (approx. USD 16.58/month) | Same as above |
| Subanana | Free | Per file | First 15 minutes preview | Export is a paid feature |
| CantoSub | Trial | New users | 30 minutes | No credit card required |
| Notta | Free | Monthly | 120 minutes, max 3 minutes per single item | — |
| Notta | Pro | Annual equivalent | Approx. USD 8.17/month, 1,800 minutes per month | — |
One more point that is easily misunderstood: With Tinrec's team version, under an active plan and active seats, live recording is currently not deducted by minute; only file and web imports consume team shared import minutes, and do not use members' personal quotas.
Conclusion: Choose by Need
Under the data conditions of this comparison, you can decide as follows:
- Need meeting workflow and team knowledge retention: Choose Tinrec, focusing on bot-free recording, post-meeting summaries and action items, team space, and seat management.
- Need quick processing of public videos or links: Subanana has the shortest workflow, but free preview only shows the first 15 minutes.
- Need specific Cantonese branches or dictionaries: CantoSub's branch labeling and dictionary settings are the most direct.
- Need clear minute-based plans: Notta's free and paid tiers are easiest to estimate.
Final reminder: this article has no standardized timing data, and the sample is insufficient to infer all scenarios. The most practical approach is to pick a real recording with mixed Chinese-English and proper nouns, run it through the free-tier versions of two tools, and the difference will be very clear.
FAQ
Q1: Is the free quota enough to judge whether a tool is good?
Most tools leave room for verification: Subanana previews the first 15 minutes per file, CantoSub gives new users 30 minutes, Notta offers 120 minutes per month, and Tinrec has a free version and a 7-day team trial. It's usually enough to test a difficult recording, but running a full one-hour meeting requires payment.
Q2: What should I look for in a tool for mixed Chinese-English meetings?
Look for explicit handling of mixed Chinese-English, not just "multilingual support." Subanana and Tinrec both explicitly mention mixed Chinese-English or mixed content organization; for other tools, it's best to test with your own recordings.
Q3: Is colloquial-to-written conversion necessary?
If the transcript will be read by someone who hasn't heard the recording, or placed in a formal document, the difference is significant. CantoSub and Subanana both list colloquial/written output as an option.
Q4: Can I demand a fixed recognition accuracy rate?
Not recommended. Accuracy is affected by recording quality, noise, accent, overlapping speech, and technical terms. When you see claims like "fixed 98%," first check whether the test conditions are close to your recording environment.
Q5: What's the difference between team and personal versions?
Personal version data belongs to the individual account. The team version is a separate space where data belongs to the team, with additional members, roles, seats, shared quotas, usage analytics, and auditing. Note that roles and seats are two different things: members without a seat can still view, play, and read team data, but cannot upload, edit, or export.
Q6: What should I be aware of when transcribing recordings?
Recording and transcription involve local regulations and participant consent; obtaining consent before meetings is safer. Also, AI transcription may still miss words or mishear; for important decisions and numbers, it's advisable to cross-check against the audio timestamps.
References
- Cantonese Speech to Text | AI Cantonese Speech Recognition | CantoSub
- Free Cantonese Speech-to-Text App 2026 | Subanana
- Cantonese Text-to-Speech
- [Free Tool] Cantonese Speech-to-Text App, Web-Based! Just Upload a File or YouTube Link. Up to 96% Accuracy - YouTuber AI | Spend time on creativity, leave the AI to do the hustle work.
- 2026 Latest 6 Cantonese Recording-to-Text AI Tools: Solving Transcript Organization and Mixed Chinese-English Pain Points - Tinrec Blog
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

4 iPhone Speech-to-Text Tools Tested: Cantonese Transcription vs. Post-Meeting Summaries—Which Saves More Time?
Compare four ways to transcribe Cantonese recordings on iPhone—built-in Voice Memos, Notely Voice, Cantonese.ai, and Tinrec—across transcription, post-meeting organization, cross-device access, and team collaboration to help you choose the right tool.

3 AI Real-Time Cantonese Translation Tools Tested for 2026: Which Offers Enough Free Quota and Post-Meeting Organization?
Starting from campus group discussions and class lectures, this article shares hands-on experience with AI real-time Cantonese translation tools. It covers 4 key purchasing criteria, Tinrec's post-meeting organization workflow, and the scenarios where OpenL Translate, Subanana, and Transync AI each fit best, helping students choose the right tool on a limited budget.

3 iPhone Cantonese Live Translation Tools Compared in 2026: Which Is Better for Meetings vs. Travel?
Starting from three common misconceptions, this comparison examines the real-world differences between iPhone's built-in live translation, Apple's Translate app, and Tinrec in Cantonese scenarios. It covers hardware requirements, language settings, and key pitfalls to avoid, and explains which tool suits which need.

2026 AirPods Live Translation Cantonese Guide: Supported Models, Setup Steps, and Meeting Notes Integration
Can AirPods Live Translation handle Cantonese? Starting from the decision points readers struggle with most, this guide covers supported models, Apple Intelligence setup steps, the actual support scope for Cantonese and Traditional Chinese, and compares AirPods Live Translation with Tinrec in a single workflow: when wearing earbuds is enough, and when you need a transcript you can keep and reprocess.

4 Free Cantonese Speech-to-Text Tools in 2026: Free Limits and Transcript Uses
A roundup of 4 free speech-to-text options that handle Cantonese, comparing free quotas, spoken-to-written conversion, speaker labeling, and subtitle export formats, plus common pitfalls and a method to verify with your own recordings. Information verified as of August 2026.

2026 Comparison of 4 Cantonese Speech-to-Text Apps: Which Saves More Time with Code-Switching and Post-Meeting Organization?
A one-hour Cantonese meeting often takes three hours to transcribe. This article reviews four tools that handle Cantonese speech-to-text, comparing them on mixed Chinese-English recognition, input sources, post-meeting organization, team data ownership, and free tiers, plus common pitfalls and scenario-based recommendations.

4 Free Cantonese Speech-to-Text Tools Tested for 2026: Which Free Tier Is Enough?
I have over a dozen meetings a week, many with Hong Kong teams in Cantonese mixed with English. This article uses a mid-level manager's real work scenarios to compare the free tier algorithms, Cantonese recognition performance, and post-meeting organization capabilities of four Cantonese speech-to-text tools: Tinrec, Subanana, UniScribe, and cSubtitle. It also summarizes three common pitfalls to help you judge whether the free version is sufficient.

How to Use iPhone Cantonese Live Translation in 2026: 5 Methods + 4 Key Settings
A guide to 5 ways to achieve Cantonese live translation on iPhone in 2026, including Apple Intelligence translation in Messages, Phone, FaceTime, and AirPods, plus Tinrec's meeting transcription and translation workflow, with details on device requirements, language settings, and practical limitations.

2026 Meeting Audio Transcription Buyer's Guide: 7 Solutions Compared and How to Choose
Meeting audio transcription isn't just about dropping a recording into a tool. This guide starts with the most common pitfalls, then reviews the positioning, quotas, and ideal users for 7 solutions—Tinrec, Otter, Notta, TurboScribe, Granola, NotebookLM, and CLOVA Note—and offers actionable selection advice to save you from re-listening to recordings after meetings.
