Video Understanding AI: Upload a Video or Audio File and Ask Risphai Anything

Risphai can now understand video and audio. Attach a video up to 10 minutes or an audio file up to 2 hours, ask your question, and Risphai answers using what it sees and what it hears, with timestamps.

Many tools that "chat with video" only read a transcript, so they miss everything that is shown but not said. Risphai looks at your video one frame per second and listens to the audio, so it can tell you what happens on screen, what is said and when both happen. It works on every plan, including Free.

What Video Understanding Looks Like

We uploaded a 12-second clip from our launch video, which shows Risphai's media generation features, and asked a simple question.

Here is Risphai's real answer to "What can you see?" about the same clip:

Risphai read the on-screen titles, described each scene at the right second and caught the single spoken line, "Hey, how are you?", at 00:09–00:10. It also noticed that the voice line belongs to the text-to-speech segment, which means it connected what was said with what was shown.

How Risphai Understands Video and Audio

When you attach a file, Risphai prepares it in a few steps that you can follow in the upload progress window:

1. It looks at the picture. Your browser takes one frame per second, each with its timestamp, plus a preview image of the first frame.
2. It listens to the sound. The audio is transcribed with timestamps. Videos without sound skip this step and its cost.
3. It checks the content first. The frames, the preview image and the transcript are checked by OpenAI's moderation service. Your video file is uploaded only after everything passes.
4. It answers with context. When you ask a question, the transcript goes to the AI model that answers, and the frames go to a model that can analyze images, such as models from OpenAI, Google, Anthropic or xAI.

Audio files work the same way without the frames: the file is transcribed with timestamps, and every model on Risphai can use the transcript. If you only want the text, the microphone button also has a separate transcription tool.

What You Can Ask

Video understanding is useful whenever you need something from a video or recording without watching or listening to all of it:

• Summaries: "Summarize this lecture in five bullet points with timestamps."
• Finding a moment: "At what time does the speaker mention the budget?"
• Scene breakdowns: "Give me a shot-by-shot breakdown of this ad."
• On-screen text: "List every title and caption that appears in this video."
• Meeting notes: "Write the decisions and action items from this recording."
• Feedback on your edits: "Where does the pacing of this video feel slow?"
• Comparing sound and picture: "Does what the narrator says match what is shown?"

Follow-up questions keep the file in context, so you can keep asking about the same video in the same conversation.

Supported Files and Limits

Video and audio uploads work on every plan, in regular and incognito chats.

Video Understanding Pricing

Transcription uses a fixed rate of 50 credits per hour of audio, calculated from the length of the file:

Videos without sound are not transcribed, so they have no transcription cost. If transcription fails, the credits are returned. If a finished transcript or the frames are then flagged by content screening, the upload is rejected and the transcription credits stay charged.

Each answer about your file uses credits like any other response: the frames and transcript are sent with your messages as input, plus the processing charge described in the Terms. The exact credits are shown for each response, and current rates are on the Information page.

The Free plan includes 100 credits per month, enough to try video understanding with short clips. A 2-hour recording uses 100 credits for transcription alone, so long recordings and regular use are a better fit for a paid plan with 3,000 to 20,000 credits per month. Compare plans on the Pricing page.

How to Use Video Understanding

1. Open a chat on Risphai and sign in.
2. Attach your file. Click + and choose a video or audio file, or drag it into the chat box.
3. Wait for the upload. Click See progress to follow each step.
4. Ask your question. For example: "What happens in this video? Give me a timestamped breakdown and tell me exactly what is said."

In Advanced mode on Paid plans, you can choose the preset or model that answers.

Good to Know

• One frame per second. Details that appear for less than a second can fall between frames, so very fast motion may be missed.
• AI can make mistakes. Check important details, especially names, numbers and exact quotes, against the original.
• Recordings of other people. Only upload recordings you are allowed to share. Risphai does not use your uploads to identify people by their face or voice.
• Files stay until you delete them. Uploaded videos and audio appear in the File Manager under Uploaded Videos and Uploaded Audios, where you can delete them.

Privacy

Audio is transcribed by xAI, and the frames, preview image and transcript are checked by OpenAI's moderation service before your video is uploaded. Neither step is covered by Zero Data Retention. Files are stored until you delete them, the conversation or your account, and files from incognito chats are scheduled for automatic deletion, as described in the Privacy Policy.

Try Video Understanding

Video and audio understanding is live on Risphai now. Attach a video or a recording, ask what you want to know, and Risphai answers with what it sees and hears. For more included credits, the higher quality presets and model selection, see the Pricing page. All changes in this release are in the v0.1.3 release notes.