Free one-sentence Mandarin speaking checker - feedback welcome

Language Learning Chinese
1

Hi everyone,

I made a small free Mandarin speaking checker for learners who want a quick check on one spoken Chinese sentence.

You record one practical sentence, and it gives feedback on pronunciation clarity, tone control, fluency, and expression quality.

Tool: free Mandarin speaking checker

A few notes so expectations are clear:

  • It is not an official HSK/HSKK score.
  • It is not a full speaking exam.
  • It is meant as a quick practice diagnostic to help you spot what to work on next.

I would really appreciate feedback from Chinese learners here, especially:

  • Does the feedback tell you what to improve next?
  • Are the four categories clear or confusing?
  • Is the scoring too generous, too harsh, or just not useful?
  • Would example recordings at different levels make the result more actionable?

If this is not a good fit for the forum, no problem. Please let me know and I will remove it or reword it. :grin:

2

No, you didn’t make a free one-sentence checker. You’re posting on behalf of a company that is selling an AI-based training product.

I first tested with my normal Chinese reading (having spoken Chinese for 40+ years). The tool felt that I need to “work on tone control” when saying “zui4 jin4”. This despite “Tone-level detail was limited in this attempt, so the report leans more on word matching and overall pacing.” In other words, the feedback isn’t very useful. It’s grasping at something, anything.

When I re-read the same sentence using a horrible foreign accent, almost no tones, I got mostly similar scores, about 2 points lower for “pronunciation clarity” and 2 points lower for “tone control”, even though there was virtually no tone control in the sample.

I’m sure you’ll get a lot of folks who are impressed with all the statistics, but I’m not convinced those statistics are really measuring anything useful.

3

Thanks for taking the time to try it and write this out. This is genuinely useful feedback.

Just to clarify one thing: the checker itself is free to use. Elo is a product/company, yes, but nobody has to use the paid product after trying the checker. The only reason there is a daily limit is that each assessment has API/processing cost on our side, so I had to put a cap on free checks.

Your criticism of the result is fair. If the tool says tone-level detail was limited, it should not then confidently tell you to work on tone control. And if a clearly worse second reading only changed the score by a couple of points, that suggests a real issue in the scoring/feedback logic, not just wording.

The goal of this tool is to be useful to learners, not to show impressive-looking numbers. I’ll check the logs for this case and review how the current assessment result is being mapped into feedback. I’ll also adjust the logic so uncertain tone evidence is handled more honestly. Thanks again for testing it seriously! :heart:

4

If you can just reword your original post to reflect the actual nature of the page you’re directing people to, that would be great. “Learning Chinese” does not permit misleading posts or one-off posts that are simply to drive traffic to a commercial site, particularly when the poster is not a regular contributor to the community.

You can also consider advertising with us…Forumosa has great coverage and many people who are not actively reading the Learning Chinese forum might be interested.

5

I’m happy to reword the original post to make the Elo/commercial context clearer upfront.

I’m trying to edit the post now, but I don’t seem to see an edit option on the original post. If editing is limited for new users or after a certain time, please let me know what I should do and I’ll update it as soon as I can.

Also, thanks for mentioning advertising. Where can I find the advertising information or the right contact/page for that? :grinning_face:

6

Don’t want to flag as it’s not really off-topic, but dunno if @ironlady can help here - @tempogain ?

(Flagging mine now though, pls temp, kthxbai!)

7

Got a 93, with the feedback:

“好你最” sounded too segmented. Say this part once as one chunk, then reconnect it to the full sentence.

Um, but there was a comma in between 你好 and 你最近？

8

Thanks for giving it a try and pointing that out.

I made a quick fix just now. If you have a minute, could you try it again and let me know how it goes? :heart:

9

Made a round of improvements to the checker.

The overall score should now be more balanced, especially when the evidence is uncertain, and the result page should also show more concrete details instead of just giving one broad score.

Would love for people to try it again and tell me if the feedback feels more useful. :hand_with_fingers_splayed: