4 Cantonese Speech-to-Text Apps Compared (2026): Free Tiers and Accuracy Explained

When choosing an app to convert Cantonese speech to text, you really only need to judge three things: how far the free tier goes, whether it can understand mixed Chinese-English, and whether you can directly use the large block of text after transcription.

This isn't a spec sheet; it's my judgment after trying them one by one.

Why the word "free" often leads to pitfalls

A 20-minute Cantonese interview typically takes over an hour to transcribe by hand. If you also need subtitles, double that time.

What's more troublesome is the Hong Kong speaking habit: mixing Chinese and English. A sentence like "個 deadline 要 push 到下個 sprint" run through a Mandarin model will likely be written as phonetically similar Chinese, and the entire transcript will need to be redone from scratch.

You may have encountered these situations:

• Advertised as free, but only gives you a few minutes of trial, and exporting costs extra.
• Transcription quality is good, but only supports English; Cantonese input fails completely.
• It's genuinely free, but you have to set up the API and adjust settings yourself—just the prep work makes you want to give up.

These aren't cases of "bad tools," but rather choosing the wrong use case from the start. Figure out what you need first, then pick a tool—it will save you a lot of time.

Before choosing a Cantonese transcription app, understand these 4 key points

1. Does the free tier cover "transcription" or "export"?

This is where most people lose out. Some tools' free plans do let you upload and see the transcription result, but as soon as you want to download the file, copy the full text, or export subtitles, a payment prompt appears. When evaluating, confirm whether the free plan gives you a "usable finished product," not just a preview.

2. How mixed Chinese-English is handled

Cantonese content almost never lacks English words. A good tool will keep the original English words and not force them into phonetically similar Chinese; more advanced ones even let you maintain your own glossary—terms like company product names or project codes will be much more stable if added in advance.

3. Is the output colloquial or written?

Sometimes a transcript needs to preserve the original colloquial flavor; other times it needs to be directly turned into presentable written Chinese. Whether you can switch, and whether you can choose Traditional or Simplified Chinese, directly affects how much time you'll spend editing later.

4. What else can you do after transcription?

This is the most easily overlooked point, but it makes the biggest difference. If you only get a transcript, you still have to read it yourself, extract key points yourself, and list action items yourself. If the tool can automatically generate summaries, chapters, and action items, or even let you ask questions directly about the content afterward, you save more than just typing time.

Think through these 4 points, and you'll realize what you should really compare is "what you get after the meeting," not whose free minutes are more.

Tinrec (秒聽錄音) — Our top pick after hands-on testing

Tinrec (秒聽錄音) is an AI meeting notes and collaboration tool for individuals and teams, supporting iOS, Android, and web, with a desktop version that can directly record online meetings. Its core isn't just "converting sound to text," but organizing the entire content into usable data.

During meetings, open Tinrec to record and get a live transcript; after finishing, AI automatically generates a summary, chapters, and action items—who is responsible for what, which tasks to follow up on, all at a glance. The desktop version works differently: it directly captures the computer's system audio to record online meetings on Zoom, Google Meet, Microsoft Teams, Webex, etc., without inviting a meeting bot to join, which is much cleaner for participants.

After recording, you can ask questions directly about the meeting content. For example, ask "Who mentioned the budget in the last meeting?" and it will give you an answer based on semantics, rather than throwing a string of keywords for you to search yourself. This AI Q&A is relatively rare among tools at the same price point. The organized content can also be turned into reports, tables, or documents, and exported to Notion, Google Docs, OneNote, Dropbox, etc. for further processing.

For mixed Chinese-English and multilingual, Tinrec covers Chinese, Traditional Chinese, and multilingual meeting scenarios, and also supports real-time translation during recording, with options to switch between original text, translation, or bilingual view. The team version also allows maintaining "hot words" to add company-specific terminology, making proper noun recognition more stable. Note that actual recognition performance is affected by recording quality, background noise, accent, multiple people talking over each other, and technical terms—same for any tool. My advice: take your own most difficult Cantonese mixed Chinese-English recording and test it once with the free tier; that's more accurate than any official numbers.

A few things I find very practical:

• No extra hardware needed; works on phone, computer, and web, no need to re-learn when switching devices.
• Recordings are retained, and the transcript has a timeline to match back to the original audio, so you can verify important content yourself.
• The team version centralizes meeting data in a shared team space; new members can see past meetings upon joining, and nothing disappears when someone leaves.

Limitations to be clear about: The free version offers basic quota, suitable for trying out recording, importing, summaries, and transcription; if you have many meetings or long interviews every week, you'll need to upgrade. The team version is currently USD 29.80 per paid seat per month, or USD 199 per paid seat per year (about USD 16.58 per month), with each paid seat providing 2,000 minutes of shared team import quota per month; eligible teams can try 7 days, 1 free seat, 300 minutes shared import quota for the first time. Actual prices and benefits are subject to the official purchase page.

Who it's for: If you frequently handle Chinese or mixed Chinese-English meetings and interviews, want to use AI to find key points right after recording, or need to preserve meeting data for team sharing, Tinrec is currently the most suitable choice.

Besides Tinrec, what other options are there?

Subanana: A Hong Kong Cantonese transcription service that lets you upload audio, video, or paste a YouTube link, producing a transcript with speaker separation and punctuation, and you can choose colloquial or written output. Free accounts can preview the first 15 minutes of each file, but export is a paid feature. If you just occasionally want to transcribe a Cantonese recording, you can try it; but it lacks team space, seat and usage management, and cannot directly record online meetings like Tinrec's desktop version.

Otter.ai: A mature choice for English meetings, with a free plan of 300 minutes per month—generous quota. However, its mindset is geared toward English business meetings; Cantonese is not its strong suit. If almost all your recordings are in English, it's more cost-effective than Tinrec; but if you need Chinese and multilingual meetings, bot-free desktop recording, and team data preservation, Tinrec is a better fit.

Google AI Studio (Gemini): A free route many recommend, with decent Cantonese support, works on desktop and mobile, and no API setup needed. Its positioning is a general AI tool; you have to copy and paste the transcript yourself, manage it yourself, and it lacks project archiving, action item extraction, subtitle export formats, and team permissions—these meeting workflows. Suitable for those with zero budget who only do one-off organization.

Pitfall guide: The 4 most common mistakes when choosing a Cantonese transcription app

Pitfall 1: Registering just because you see the word "free." Many services' free tier is "preview-based"—for example, you can see the first 15 minutes, but export requires payment. First confirm whether the finished product you want (transcript, subtitle file) is available in the free plan, then decide whether to spend time uploading.

Pitfall 2: Only looking at officially published accuracy rates. Accuracy numbers without testing conditions are meaningless. A quiet recording studio and a real meeting room differ significantly: air conditioning noise, keyboard sounds, and multiple people talking over each other will degrade performance. The most reliable approach is to test once for free with your own recording, not just look at ads.

Pitfall 3: Ignoring mixed Chinese-English and output formats. If English words in Cantonese content are written as phonetically similar Chinese, editing the transcript may take longer than retyping it. Also, decide upfront whether you want a colloquial transcript or written language, and whether you need subtitle formats like SRT or VTT.

Pitfall 4: Treating it only as a typewriter. If you only use a transcription tool as "sound to text," you're only using part of its capability. For example, Tinrec lets you ask directly after recording, "What was the conclusion of this meeting?" or "Who promised to do what?" without having to flip through a 30-minute transcript yourself. For team use, also ask: whose account is the data stored in? Can it be retrieved after a member leaves?

Conclusion: Which one should you choose?

• Need meeting notes in Chinese, Traditional Chinese, and mixed Chinese-English, and want automatic summaries and action items after the meeting → Tinrec
• Want to directly record online meetings on Zoom, Meet, Teams, etc., without a bot joining → Tinrec (desktop version captures system audio)
• Team needs to share meeting data and manage members, seats, and usage → Tinrec Team Version
• Need to follow up on transcripts with questions and export to Notion or Google Docs for further processing → Tinrec
• Zero budget, only occasionally organize a Cantonese recording → Subanana's free preview or Google AI Studio can be tried first
• Almost only record English meetings → Otter.ai's free quota is plenty

My advice: don't rush to pay. Use the free tier to upload one of your most challenging Cantonese recordings, see if the transcript, summary, and action items can be directly put to use, then decide whether to upgrade—that's the least regretful approach.

References

• Free Cantonese Speech-to-Text App 2026 | Subanana
• Speech to Text - Google Play App
• [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.
• "Free + Cantonese OK: How to Use Google AI Studio to Convert Your Recording to 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 lot after a few minutes of trial, or only supporting English, not Cantonese, or requiring a whole API setup to use. // Recently I found that Google Gemini is world-class in this area: - [Truly Free] No subscription fee after a few uses - [Cantonese-friendly] Gemini accurate
• Free! 2024 TOP3 Text-to-Speech Tools | Supports Cantonese & Mandarin & Guoyu (Taiwan) | Can Generate Celebrity/Cartoon Voices - YouTube