Why Does the Same Person Get Slightly Different Results Each Time?
The algorithm analyzes the current recording. Wording, tone, emotion, distance, microphone, noise reduction, and environment can all change the pitch trajectory and spectrum. Small changes are normal; if results differ substantially, check recording quality and the device first.
Can the Results Determine My Actual Sex or Age?
No. Feminine, masculine, and neutral voice tendencies describe perceptual distributions, and perceived age is also an impression of expression. They cannot confirm biological information and must not be used for identity verification, job screening, medical decisions, or other high-stakes purposes.
How Are Timbre Names Chosen?
The system matches multiple acoustic dimensions against each label’s target range, then selects higher-matching combinations from a candidate pool consistent with the current voice tendency. The names are meant to help describe how the voice sounds; they are not human ratings, and there is no single uniquely correct answer.
Why Was My Audio Detected as a Special Sound?
When audio is closer to whistling, sustained humming, or another non-speech vocalization, ordinary voice labels do not have a reliable basis. The system returns a separate special-sound result and keeps it out of ordinary gender/age impressions, primary-timbre statistics, and EchoSea voice-star data.
Are Recordings Stored or Used for Training?
Raw recordings in CVoice and Insight are used only to complete the current analysis and are deleted after completion or failure. They are not submitted to third-party AI systems for training. Anonymous statistics retain only classifications and counts. Audio publicly posted to EchoSea is a separate user-initiated flow and is stored and expired according to that product’s rules.
Why Can’t Some Android or In-App Browsers Record, Upload, or Verify?
Browsers differ in their support for recording codecs, file selection, cookies, human verification, and system permissions. First check microphone and file permissions, then try another browser environment such as Chrome, Edge, Safari, QQ, or WeChat. The embedded web container in desktop QQ may be unable to complete Cloudflare Turnstile; if that happens, copy the link into a regular browser.
Why Is the Quality Still Low After Uploading AMR or M4A?
The service can convert these formats into audio the algorithm can read, but it cannot restore detail already removed by compression or noise reduction in the recorder. AMR is usually optimized for voice communication rather than preserving full timbre. Successful decoding does not mean it has the same quality as a lossless recording.
Does CVoice Cost Money?
Basic experiences are provided as described on each app page. Support through Afdian is voluntary and goes toward servers, networking, and maintenance; some clearly labeled priority uses or additional experiences may correspond to specific benefits. Ordinary sponsorship does not change test results.
Why Does the Site Show Ads?
Ads help cover server and maintenance costs. They do not participate in the algorithm and should not be disguised as action buttons. If an ad interferes with reading or contains inappropriate content, you can notify the developer through page feedback.
How Do I Report an Error or Inappropriate Content?
Use the feedback or report entry points in each app, or join QQ group 1087037352. Please include the device, browser, time of occurrence, and reproducible steps. If privacy is involved, do not publicly send recordings, tokens, or other sensitive information.