Onko AI

What is Onko AI?

Yinzi AI is an online AI processing platform for audio and video, developed by Guangzhou Youyidian Intelligent Technology.

It offers track separation, text-to-speech conversion, AI voice synthesis, and short video analysis, and provides APIs.

Core functions

Track separation

Track separation allows the vocals and accompaniment in audio or video to be separated into separate files.

• It is suitable for creating cover song accompaniments, extracting dialogue, and performing post-production mixing.
• The separation quality is affected by compression, reverb, harmony, and background noise.
• Try to use original files that are clear and have a high bitrate before processing.
• After downloading, check for issues such as voice leakage, loss of musical instruments, and phase problems.
• Professional distribution still requires the use of an audio workstation for manual editing.

Voice and accompaniment extraction

Users can extract only the vocal track or keep only the accompaniment, thereby reducing the number of unnecessary output files.

• Voice extraction is suitable for interviews, dialogues, podcasts, and a cappella material.
• Accompaniment tracks are suitable for singing practice, performances, and as background music for videos.
• Choral singing by multiple people and a large number of instruments can make it more difficult to achieve separation.
• The copyright of the material does not change as a result of technical processing.
• Authorization for the recording, lyrics, music, and performance must be obtained before making them public or using them for commercial purposes.

Speech to text

The script extraction feature can convert the speech in audio or video into text.

• It is suitable for organizing subtitles, interviews, courses, podcasts, and meeting minutes.
• Dialects, overlapping speech from multiple speakers, noise, and proper nouns may reduce accuracy.
• After generation, the numbers, names, dates, and technical terms should be checked.
• Before uploading sensitive recordings, it is necessary to obtain the participants’ permission and remove any sensitive information from them.

Text to speech

The text-to-speech function supports multiple languages and various voice tones, and it allows for adjusting the speech speed and pitch.

• You can paste text directly or upload a text file.
• First, listen to different speakers to find the voice that suits the piece.
• Suitable for audiobooks, lesson explanations, news broadcasts, and video dubbing.
• After generation, check for polyphonic characters, pauses, English abbreviations, and the way numbers are pronounced.
• It is forbidden to impersonate real people or to clone recognizable voices without permission.

Analysis of short videos

The open platform can parse information related to short video shares and return the video and background music associated with them.

• The API documentation lists platforms such as TikTok, Kuaishou, REDnote, and Weibo.
• Only one’s own works or content for which permission to download and reuse has been obtained should be processed.
• Removing the platform watermark does not equate to obtaining copyright or commercial rights for the work.
• In the event of a parsing failure, the official billing page states that no points will be deducted.

Guide to Using Onko AI

1. Open the corresponding tool and select track separation, text conversion, or voice-over functionality.
2. Upload audio, video, or text materials for which you have permission to handle.
3. Follow the on-screen instructions to confirm the file length, format, and estimated points.
4. For the text-to-speech task, you first need to select the language, voice tone, speaking speed, and pitch.
5. Submit the task and wait for it to be processed in the cloud.
6. Preview the result online to check for issues with sound quality, subtitles, or pronunciation.
7. Download the result and carry out editing, proofreading, or post-production work locally.
8. Delete the sensitive materials that are no longer needed, and keep a backup of the original files.

Price of the subscription package

The following is the publicly available pricing information in Simplified Chinese as of August 30, 2026.

The currency used and the package prices on pages for different regions may vary; it is necessary to check the actual payment location and the settlement page for details.

Prices, bonus points, and functional rules may change; the terms applicable are those stated on the purchase page, the checkout page, and the current agreement.

Functional billing rules

The page for extracting copy text provides different guidelines based on the number of characters and the duration of the media content; before submitting, it is necessary to follow the estimated fee indicated by the tool in use.

Validity period of points and refunds

• The points purchased are valid indefinitely, with no set expiration date.
• When the balance is insufficient, it is necessary to purchase any package to top up points.
• The processing failed due to system issues; according to the official statement, no points will be deducted.
• Refunds are generally not available after the purchase using points.
• For special cases, an application must be submitted through customer service for processing.

Open API

Onson AI offers a v1 open platform API, which uses an API Key generated in the personal account for authentication.

• Create an asynchronous order processing and obtain the order number.
• Query the processing status, point cost, and result files.
• It supports track separation, vocal extraction, accompaniment extraction, and text extraction.
• It supports real-time parsing of short video sharing information.
• As a result, the download link is usually valid only for a limited time.
• When the web version does not support batch uploading, submissions can be made in batches via API.

API Keys are sensitive credentials and should not be included in frontend code, published in repositories, or shared in files.

API package options and calling considerations

• The API user package costs 1,000 yuan and provides 50,000 points.
• Asynchronous tasks require polling the order status; it is not sufficient to merely check whether submission was successful.
• A minimum number of points is reserved when an order is created, and the settlement is made based on the actual usage once the order is completed.
• Logic for handling failures, timeouts, duplicate submissions, and insufficient balance should be established.
• Save the downloaded result promptly to prevent the temporary link from expiring.

Open-source status

• Product source code: As of the time of verification, no official public repository was found.
• Model weights: No official publicly available weights for download were found.
• SDK: HTTP API documentation is available, but no separate official open-source SDK has been found.
• Private deployment: No public solutions available for ordinary users have been found.

An open API merely means that it is possible to call the services; it does not imply that the Yinzai AI products or models are open-source projects.

Usage restrictions and precautions

• AI separation may result in residual human voice, missing accompaniment, and reduced sound quality.
• Speech recognition and voice synthesis may result in spelling errors, incorrect pauses, or unnatural pronunciation.
• It is not allowed to process unauthorized videos, music, or private recordings of others.
• The short-video analysis feature must comply with the platform’s rules and the copyright regulations regarding the content.
• Before commercial release, it is necessary to verify the licensing scope of the generated content, sound tones, and materials.

Frequently Asked Questions

What can Yoonzi AI do mainly?

It can separate audio tracks, isolate vocals from accompaniment, convert speech to text, and analyze short videos.

Is there a free trial for Onko AI?

Yes, the same user or device can earn 50 points at a time; no registration is required to try it out.

How is charging handled for track separation?

30 points are deducted within 3 minutes; thereafter, an additional 15 points are deducted for each extra minute.

Will the purchased points expire?

No, the official pricing page states that the purchase credits remain valid indefinitely, until they are used up.

Does Onson AI support APIs?

Supported: it allows for the creation and retrieval of orders, as well as the use of functions related to audio tracks, vocals, accompaniments, text, and short videos.

Will points be deducted in case of failed processing?

Points will not be deducted in the event of failure due to system-related issues; the order records and current rules shall prevail.

Is Onko AI an open-source project?

No, no official source code or open weights have been found; providing an API does not mean that the product is open-source.