This comparison is published by TonePerfect. Competitor facts were checked on 10 September 2026 against the vendor's public pages linked below; prices and features change, so confirm on their site before you decide. Where the other product is the better choice, we say so.

Where SpeechSuper is the better choice

  • More languages. SpeechSuper lists eight: English (with US, British and Indian variants), Mandarin, German, French, Spanish, Korean, Japanese and Russian. TonePerfect does not serve Korean, Japanese or Russian.
  • Unscripted and exam-style assessment. SpeechSuper offers semi-scripted and unscripted English products (IELTS and PTE-style speaking, topic development, grammar and vocabulary). TonePerfect assesses scripted speech only: the learner reads a sentence you supply.
  • Offline mobile SDK. SpeechSuper licenses an iOS/Android SDK that runs on device (from $1 per device per year at 2,000+ devices, checked 10 September 2026). TonePerfect is server-side only.
  • Longer inputs. SpeechSuper has a paragraph product for texts over 200 words. TonePerfect accepts up to 30 seconds per request.
  • Track record. SpeechSuper has served edtech customers for years; TonePerfect's language packs are labelled experimental in /v1/capabilities while calibration data grows.

Where TonePerfect is the better choice

  • Mandarin syllable structure. TonePerfect scores the initial, final and tone of every syllable as separate units, each with an expected value, an estimated heard value and a score. A learner who says shì with a 2nd tone sees "Tone 4 → heard Tone 2", not just a low syllable score. Third-tone sandhi is taken into account so correct 3-3 → 2-3 speech is not marked as an error.
  • One response shape for every language. The same scores → words[] → phones[] JSON covers all seven languages, so a multi-language app has one integration and one UI.
  • Correction decisions, not just numbers. Every Mandarin component and every word in the other languages carries a decision — review, uncertain or no_correction — which separates confident errors from uncertain evidence so your app can decide when to stay quiet. See how scoring works.
  • Simpler pricing. Flat monthly plans, one credit per recording up to 15 seconds, no quarterly invoicing and no $20 monthly minimum. Failed or unassessable requests cost nothing.
  • Privacy by default. Audio is deleted after scoring and never used for training; results are kept for at most 24 hours to make retries idempotent.

Side by side

FeatureSpeechSuperTonePerfect
LanguagesEnglish (US/UK/IN), Mandarin, German, French, Spanish, Korean, Japanese, RussianMandarin Chinese (zh-CN), English (en-US), French (fr-FR), Spanish (es-419), German (de-DE), Italian (it-IT), Portuguese (pt-BR)
Scripted assessment levelsPhoneme, syllable, word, sentence; pronunciation, fluency, completeness, rhythmSound, word, sentence; pronunciation plus completeness, fluency and rhythm where the pack supports them; Mandarin initial / final / tone
Unscripted / exam scoringYes — semi-scripted and unscripted English (IELTS, PTE), Mandarin speech assessmentNo — scripted practice only
Mandarin tone feedbackPer-syllable scoresPer-syllable initial, final and tone with expected vs estimated heard tone and a decision; sandhi-aware
Expected vs heard soundMispronunciation detectionExpected and estimated heard value for every unit; decisions per Mandarin component and per word in other languages
Word timestampsNot verifiedYes (start_ms, end_ms)
Audio length per requestWord, sentence (2–200 words) and paragraph (200+ words) productsUp to 30 seconds, 10 MB
DeliveryREST API + offline iOS/Android SDKREST API (multipart upload)
Pricing modelPer request: word $0.004, sentence $0.006, paragraph $0.008; $20/month minimum, quarterly invoicing; prepaid tiers 15–30% off from $500Monthly plans $10 / $50 / $150 with 2,500 / 15,000 / 50,000 credits; extra packs 10–20% off; no minimum spend beyond the plan
Free trialYesNo free credits; account, keys and docs are free
Idempotent retriesNot documentedYes — Idempotency-Key header
Audio retentionSee vendor policyDeleted after scoring; never used for training

What it costs at three volumes

Monthly volume (recordings ≤ 15 s)SpeechSuper (sentence API)TonePerfect
1,000$20 (monthly minimum; 1,000 × $0.006 = $6)$10 — Starter plan (2,500 credits included)
10,000$60 pay-as-you-go$50 — Builder plan (15,000 credits included)
100,000$600 pay-as-you-go, or about $420 on the Growth prepaid tier ($0.0042)$310 — Scale plan $150 (50,000 credits) + two 25,000-credit packs at $80 each (20% plan discount)

Assumes one short sentence per recording and SpeechSuper list prices checked 10 September 2026. SpeechSuper bills per request regardless of length within a product; TonePerfect charges two credits for recordings between 15 and 30 seconds. Unused TonePerfect credits roll over while the subscription is active.

Integration differences worth knowing

SpeechSuper's request format is a JSON "coreType" parameter set (for example sent.eval or word.eval.cn) sent with the audio; each language and task has its own core and its own response fields. TonePerfect has one endpoint, POST /v1/assess, that takes audio, text and language as multipart form fields and returns the same structure for every language:

cURL
curl https://api.toneperfect.app/v1/assess \
  -H "Authorization: Bearer $TONEPERFECT_API_KEY" \
  -H "Idempotency-Key: lesson-42-attempt-1" \
  -F "audio=@take.wav" \
  -F "text=我想喝咖啡" \
  -F "language=zh-CN" \
  -F "task=connected_speech"

If you are migrating, the mapping is straightforward: SpeechSuper's overall score maps to scores.overall, word scores to words[].score, and phoneme detail to words[].phones[]. The one thing you gain — and have to design for — is the decision field: instead of applying your own 40/60/70 thresholds to a number, you can let the API tell you which units deserve a correction. The integration tutorial shows both approaches.

Bottom line

Choose SpeechSuper if you need Korean, Japanese or Russian, unscripted or exam-style scoring, an on-device SDK, or long passages. Choose TonePerfect if your product teaches scripted speaking practice in one of its seven languages, if Mandarin tones matter to you, or if you want one response shape, simple monthly pricing and no audio retention. Many teams run both through the same recordings before deciding; a TonePerfect key takes a minute to create.

Sources checked 10 September 2026: SpeechSuper pricing · SpeechSuper product pages.