Input Formats and Limits
This page documents the supported input formats and processing limits for Valossa AI video analysis.
Video Format Support
Valossa AI supports most common video container formats and codecs, including but not limited to:
| Format | Notes |
|---|---|
| MP4 | Recommended. H.264 codec is fully supported. |
| MPEG | Supported |
| AVI | Supported |
| FLV | Supported |
| WebM | Supported |
MP4 with the H.264 codec is the most reliable and widely tested combination. If you encounter issues with another format, try re-encoding to H.264 MP4.
Processing Limits
| Parameter | Limit |
|---|---|
| Maximum file size | 7 GB per video file |
| Maximum duration | 5 hours of playback time |
| Maximum vertical resolution | 4096 pixels |
| Transcript file size | 5 MB per SRT file |
Video Stream Behavior
- If the video file contains multiple video streams, only the first stream is analyzed.
- If the video file contains multiple audio streams, only the first stream is analyzed.
- The audio stream can be either mono or stereo.
Video Delivery Methods
You can provide video content to Valossa AI in three ways:
1. URL Download
Provide a direct link to a video file in the url parameter of your new_job API call. The URL must point directly to a video file -- Valossa will not scrape or extract videos from HTML pages.
Supported URL protocols:
- HTTPS / HTTP -- Standard web URLs
- AWS S3 -- Can be configured for your account (contact Valossa support to enable S3 download support)
- Google Drive -- Supported if sharing settings allow link access
- Dropbox -- Supported if sharing settings allow link access
- The system downloads files asynchronously. There is a delay between job creation and download initiation.
- Avoid expiring URLs. If you must use them, set a long expiration time.
- Do not submit an unreasonable number of files per day.
2. Upload via Valossa Portal
Use the Analyze page in Valossa Portal to upload a video file through the web interface. This is the easiest method for manual, one-off analyses.
3. Upload via API
Use the uploading-related API functions to upload a file, then reference the generated valossaupload:// URL in your new_job request. See the API reference documentation for details on the upload workflow.
Transcript Input (SRT)
You can optionally provide a pre-existing speech transcript in SRT format alongside your video. This is useful when you already have accurate subtitles or transcriptions.
How Transcript Input Affects Analysis
| Scenario | Audio Speech-to-Text | Audio Keywords | Transcript Keywords | Audio Context |
|---|---|---|---|---|
| No transcript provided | Generated automatically | Generated from speech | Not available | Always generated |
| SRT transcript provided | Not generated (your transcript is used) | Not generated | Generated from your transcript | Always generated |
Key Considerations
- The only supported format for pre-existing transcripts is SRT.
- The only supported language for pre-existing transcript input is English.
- If your pre-existing transcript is provided, analysis results will be more restricted (no audio-based speech detection is performed).
- Results derived from your transcript appear in
transcript.*detection types instead ofaudio.*detection types. - Transcript-only analysis (SRT without a video file) is also supported with special configuration.
When to Provide a Transcript
Provide a transcript when:
- You already have accurate, professionally created subtitles
- You need the speech content to be exact (e.g., legal or compliance use cases)
Skip the transcript when:
- You want the full range of audio analysis features
- You do not have an existing transcript
- You want automatic speech-to-text generation