User Guide & Documentation
Complete reference for all four AI metadata generators — Free Image, Pro Image, Pro Video, Pro Audio
1. Free Image — Puter Account, Sign-in and Limits
The "☁ Puter Account" block on /img-free manages your connection to the AI provider.
- Guest access: technically you can start without pressing "Sign in / Register" — but the "Start Generation" button stays locked in that state, so no result can actually be generated. Signing in to Puter is required to generate anything, though registration itself is free and takes under a minute.
- Registration: the "🪪 Sign in / Register" button opens the Puter.com sign-in/sign-up window. A registered account gets its own monthly AI quota that persists between visits (unlike a guest session, which doesn't carry quota over).
- 🔄 Refresh account: manually refreshes your sign-in status and remaining quota.
- Quota indicator: shows monthly usage (percentage and absolute numbers), remaining quota, and the total monthly limit. The bar's color shifts as you spend it: green → yellow (70%+ used) → red (90%+ used).
- 📊 Usage Statistics: a detailed breakdown of images processed and quota spent per model for the current month, plus an estimated "Images remaining" figure per model.
2. Pro (Image / Video / Audio) — Credits VB, Token, and Balance
All three Pro tools share one prepaid-credit system instead of a Puter account.
2.1. Buying credits
The "Shop" link in the site menu, or the "Buy More Credits VB" button on any Pro page, opens the credits shop. One Credits VB package works across all three Pro tools at once — spend it on Image, Video, and Audio in any proportion.
2.2. Email and activation
Within a few minutes of payment, an email with your access token and an activation link arrives at the address you used at checkout. Two ways to activate:
- Via the link — clicking the email link saves the token to your browser automatically.
- Manually — paste the token into the "Paste your access token" field on the balance panel and click "Sign In".
The token isn't tied to a device — activate it the same way on any other browser/computer and your balance and history follow. Lost the email? Use "Lost your access token? Resend it" (same email as at checkout).
2.3. The balance panel
Shows your current Credits VB balance, plus "Buy More Credits VB" and "Export/Import Settings" (copies custom platforms/CSV presets/synced settings onto a new token — does not move balance or history, which stay on the original token).
2.4. Running-low email
When about 20% of your package is left, we email you once, to the address you used at checkout. It's a heads-up so you don't run out mid-batch — nothing is blocked, and your token keeps working normally until it's actually empty. The email never contains your access token: you already have it, and an extra copy sitting in a mailbox is just one more way to lose it.
3. How Credits Are Charged
- Charged before the AI runs, at the price shown to you in advance (on the file card and in the summary above "Start Generation"). The final charge never differs from the price you saw.
- The price is built from the limits you set in Settings (Title/Description/Keywords), not from how long the answer turns out. That is what makes it possible to show you the exact price before anything runs. Short source material gives a shorter result at the same price, so if you do not need a long description on a platform that allows a very long one, lower that limit — the price drops with it, live, before you press Generate.
- A generation that fails to produce usable metadata — a network or AI-service error, an empty response, or a reply we can't read as metadata — refunds the charged credits to your balance automatically and instantly, with no support request needed. The error message on the card tells you exactly how many credits came back and what your balance is now, so you never have to take our word for it.
- Credits VB never expire.
- Your balance may end up with a small leftover that isn't enough for any generation (e.g. 3 credits when the cheapest option costs 5) — that's not a technical fault, and that leftover isn't refunded as money. If it matters to you, spend a token down close to zero before buying the next package.
4. File and Batch Limits — a Recommendation, Not a Guarantee
The maximum file count and maximum file size (see the tables in each generator's guide below) are a technical ceiling the site allows, not a guarantee that your browser can handle a batch that size. Actual batch capacity depends on your computer's memory and performance — the browser is under the most load during preview and bulk-export steps (ZIP, embedding metadata across a whole batch).
Recommendation: not sure your computer can handle a big batch? Start with 5-10 files and increase gradually, only once you've confirmed the previous batch went through without issues. A few smaller batches beat a frozen browser tab in the middle of a big one.
5. Support
Questions, technical issues, feature ideas — through the in-app feedback form (available once you've activated a Pro token, or signed in via Puter on Free).
1. Generation Settings
1.1. AI Model
Dropdown of available models. Models differ in cost (Puter quota on Free, Credits VB on Pro) and processing quality/speed. On Free, check your remaining quota per model via "📊 Usage Statistics" (see Getting Started, section 1). On Pro, picking a model immediately shows its Credits VB price for one image at your current settings.
1.2. Platform
Choosing a target platform sets the prompt (that platform's specific requirements) and default field limits (adjustable by hand). 12 profiles total: 6 for specific platforms (section 4) and 6 universal ones by content type (section 3).
1.3. Title & Description Max Chars
Numeric max-length fields, each with its own toggle:
- Enabled — the AI generates the field within the given limit.
- Disabled — the field isn't requested from the AI at all (saves quota/credits), stays blank for manual entry or for platforms that don't need it (Vecteezy has Description off by default, for example).
Switching platforms adjusts limits and toggle states automatically to match the profile, but you can always fine-tune manually — changes are saved in your browser.
1.4. Keywords Count
How many keywords to request from the model. Defaults follow the platform's own rules; freely adjustable.
1.5. Top Keywords Count
How many of the first keywords in the list count as most important for the platform's search ranking — passed to the model as a priority instruction (main subject → action → details → environment → category → general concepts). Those same first N words are visually badged with a rank number in the keyword chip editor.
1.6. 🔑 Keyword Rules — Your Own Overrides
- Include Keywords — terms that must always be in the result. Inserted right after the "Top N" slot so limits or de-duplication can't accidentally drop them.
- Exclude Keywords (stop-list) — your personal blacklist: stripped from the result regardless of what the AI generated.
- Trademarks — a brand-name list ("iPhone", "Nike", "Coca-Cola") to auto-remove. Pre-filled with popular brands, editable for your niche.
- Keyword Format — Auto (platform default) / single words only / allow 2-word phrases.
- Auto-apply Rules — applies the stop-list, trademarks, and Include list right after every generation. Order: Exclude → Trademarks → Include, so an Include word can never be accidentally stripped by your own stop-list or trademark list.
- Auto-remove Duplicates — strips repeats (case-insensitive) right after generation.
2. Custom Platforms (Pro)
Up to 5 saved profiles for images. Each is a name + your own instructions for the AI + (optionally) your own category list. Fully replaces the preset platform once the "Use Custom Platform" toggle next to Platform is on.
3. Universal Profiles
General-purpose profiles for personal archives, cases with similar requirements across platforms, or content types that don't fit a specific-platform profile:
| Profile | Purpose |
|---|---|
| 🌐 Universal Basic | Light, neutral profile with minimal constraints — a quick start. |
| 🌐 Universal Strict | Stricter: an expanded marketing stop-list, conditional logic (no emotion/travel words without clear visual proof), tight control over semantic duplicates. |
| 🌐 Universal Photo | For photography: emphasizes the physical subject, materials, lighting, camera angle. Filters out vector/render terminology. |
| 🌐 Universal Vector/Illustration | For vector graphics and illustrations: asset type is guaranteed within the first 5 keywords, accounts for style/technique. |
| 🌐 Universal 3D Render | For 3D renders and CGI: prioritizes materials, lighting, composition. Excludes photography terms in favor of "3d render", "cgi", etc. |
| 🌐 Universal Pattern/Texture | For seamless patterns and textures: pattern type is required in the leading keywords, tracks motif/shape, palette, repeat structure. |
4. Platform-Specific Profiles
Tuned to official microstock guidelines (rules checked as of mid-2026):
| Platform | Title / Description | Keywords | Notable rules |
|---|---|---|---|
| 🅐 Adobe Stock | Title ≤ 200 (aim 70) · Desc ≤ 200 | Exactly 49 | First 10 words set rank; strict hierarchy (subject → action → environment → general concepts). Zero tolerance for brands, camera models, subjective words. |
| 🔴 Shutterstock | Title min. 5 words (≤ 150) · Desc ≤ 250 | 50 (min. 7) | Anti-spam: title and description can't repeat the same words/roots. Auto editorial format ("City, Country — Date: Description"). Prefers single words. |
| 📷 Getty / iStock | Title ≤ 180 (aim 55-70) · Desc ≤ 250 | 50 | Mandatory people-count tag ("no people", "one person", etc.), demographics only when obvious. Strict controlled vocabulary. |
| 🟢 Magnific (Freepik) | Title ≤ 100 · Desc ≤ 250 | 50 (15-20 core) | Auto-detects asset type (photo/vector/3D/pattern/AI). Filters out file extensions (EPS10, JPG, PSD), hashtags, filler words ("cool", "nice"). |
| 🟤 Dreamstime | Title ≤ 80 (aim 60-80) · Desc ≤ 250 | 7-50 | Rejects submissions under 7 keywords. Strict ban on repeating a word root ("run, running, runner" — keep one form only). |
| 💛 Vecteezy | Title ≤ 200 (aim 3-8 words) · Desc off by default | 50 | Mandatory categories when people are present: count, age, ethnicity; mandatory orientation, background; asset type within the first 10 tags. No slang. |
5. Upload and Generation
5.1. Uploading files
Drag files into the dropzone or click it to browse. Supports JPEG, PNG, WebP, GIF, TIFF, BMP, SVG, EPS, and AI. Batch limit: up to 50 images on Free, up to 500 on Pro (see also "File and Batch Limits" in Getting Started — a technical ceiling, not a guarantee for every computer). Unsupported files are rejected automatically with an explanation; files beyond the limit are simply not added.
5.2. Processing
"Start Generation" runs the queue sequentially with adaptive pacing: pauses grow if the provider is rate-limiting, shrink during stable operation. A progress counter shows "N / M".
For each image:
- Resized locally in browser memory (no server upload) to the size the model needs — saves quota/credits without losing detail.
- Sent to the model with platform-specific prompt instructions.
- The response is parsed and post-processed: the platform's stop-list, your personal Include/Exclude/Trademarks, duplicate removal, orientation tags.
- A summary under the Keywords field shows what was auto-adjusted (words removed as stop-list/trademark, words added from Include, duplicates cleared).
Errors from the provider are classified and handled differently:
- Temporary (rate limits, network errors) — the affected file retries automatically, up to 3 attempts with growing pauses. You can also retry a single card by hand with "Generate again".
- Blocking (Pro: balance ran out or token invalid; Free: Puter quota exhausted or session expired) — retrying won't help, so the whole batch pauses instead. A banner shows how many files were processed so far, with "▶ Resume" (continue once the issue is resolved) and "✕ Cancel & keep results" (stop, keeping everything already generated).
6. Result Card Interface
6.1. Title & Description
Auto-resizing text fields with a live character counter (green = within limit, orange = over) and a one-click copy button.
6.2. Interactive keyword chip editor
- Add/remove — type and press Enter or comma (or paste a comma-separated list). Click
×or press Backspace on an empty field to delete. - Inline edit — double-click a chip to edit its text.
- Reorder — drag with the mouse.
- Visual cues — Top N rank badge, yellow outline (duplicate warning), red (stop-list/trademark), green (Include word).
6.3. Card tools
A menu above the Keywords field with targeted actions for that card:
- 🏷️ Insert include keywords — insert (or top up) missing Include Keywords terms.
- 🔁 Remove duplicates — strip repeats.
- ✂ Split into words — break phrases into single words.
- 📏 Trim to limit — cut the list down to a set count.
- 🚫 Remove trademarks — strip Trademarks-list words.
- 🛑 Remove stop-list words — strip your personal stop-list words.
Each action shows a short result below the field ("Removed: 3 duplicates" or "No trademarks found").
6.4. Global history (Undo/Redo)
Every keyword edit — via the chip editor, card tools, or Batch Tools — is saved in one page-wide history. Ctrl+Z undoes the last action, Ctrl+Y (or Ctrl+Shift+Z) redoes it. Bulk edits across multiple cards undo in a single click.
6.5. 💾 Save with EXIF/XMP
Saves the image with Title/Description/Keywords embedded directly into the file — no extra credits. JPEG gets EXIF, XMP and IPTC; PNG and WebP get XMP; SVG gets native <title>/<desc> plus Dublin Core; AI gets the PDF Info dictionary plus XMP. Not available for GIF, TIFF, BMP or EPS — use CSV Export (section 8) for those. Whether the metadata already in your file (camera data, copyright, existing tags) is kept or replaced is controlled by Existing metadata in Settings.
7. 🧰 Batch Tools — Bulk Processing (Pro)
A side panel with the same actions as section 6.3 (Insert include keywords, Remove duplicates, Split into words, Trim to limit, Remove trademarks, Remove stop-list words), applied to every completed card at once with a single click. Afterward you see "Applied to N of M images"; the whole action undoes with one Ctrl+Z.
The panel stays clickable while a batch is still generating, but instead of applying the action it shows a short notice that batch tools will be available once generation finishes — clear feedback instead of a silently disabled button.
8. 📄 CSV Export — Bulk Table Export
A dedicated panel for configuring and exporting results to CSV — the format most microstock platforms accept for bulk metadata uploads. Available on both Free and Pro.
8.1. Platform presets
- 🌐 Universal (basic) — Filename, Title, Description, Keywords, Categories.
- 🅐 Adobe Stock — Filename, Title, Keywords, Category, Releases.
- 🔴 Shutterstock — Filename, Description, Keywords, Categories, Illustration, Mature Content, Editorial.
- 🟢 Magnific (Freepik) — Filename, Title, Keywords, Prompt, Model (delimiter
;, as the platform requires). - 🟤 Dreamstime — Filename, Title, Description, Keywords, Category 1/2/3, Editorial, Free.
- 💛 Vecteezy — Filename, Title, Description, Keywords, License.
- 📷 Getty/iStock — same structure as Universal (the platform has no official bulk-upload format; provided for personal use).
- 🎞️ Pond5 — OriginalFilename, Title, Keywords (Pond5 accepts image content through the same bulk importer it uses for video).
8.2. Custom column builder
Beyond the presets, a full catalog of 20+ fields — each toggled on/off and freely reordered:
- Auto-filled from generation: Filename, Title, Description, Keywords.
- Static text across the batch: Copyright, Creator, Email, Website, Country, License, AI Model, etc. — entered once, applied to every row.
- Yes/No toggles across the batch: Editorial, Illustration, Mature Content, Free.
- Empty placeholders: Category, Prompt, and similar fields that need per-image filling — left blank, marked "will be inserted empty — fill in manually" so it's clearly intentional, not an error.
Any manual change to the field set switches the preset to "🛠 Custom" — the original preset isn't lost and can be restored from the dropdown at any time.
8.3. Running the export
"📄 Export CSV" builds the file from the current values (including any manual card edits) for title/description/keywords, the right column delimiter for the chosen preset, and UTF-8 with BOM (so Cyrillic renders correctly in Excel). Downloads as stock_metadata_<preset>_<date_time>.csv.
9. Bulk Export (Pro)
- 📦 Download All as ZIP — packs every successfully processed file into one
.zip: each image is saved with metadata already embedded (the same mechanism as "💾 Save with EXIF/XMP" on a single card, section 6.5). A progress indicator shows how many files are packed. If two files would end up with the same output name, a numeric suffix (_1,_2) is added automatically so nothing gets silently overwritten inside the archive. - Save All to Folder (in supporting browsers) — the same thing, but saves straight to a folder on disk instead of an archive.
10. Generation History (Pro, /pro-history)
Off by default — turn it on in Settings. Stores filename, platform, model, and every result field for each generation. You can view it, re-download as CSV, or delete individual entries or all of it.
11. Usage Tips & Recommendations
- Set up Trademarks and the stop-list once for your shooting niche, so you don't clean brands and unwanted words out of every result by hand.
- Use Include Keywords for recurring themes — great for series/collection names that repeat across images, so you don't retype them for every image.
- Keep Top Keywords Count around 5-10 — most platforms weigh the first keywords heavily in search ranking; too high a value dilutes priority.
- Turn off Description where a platform doesn't need it (Vecteezy, for example) — matches the platform's requirements and saves quota/credits, since the field isn't requested from the model at all.
- Always check before a bulk export — the chip editor's highlights (duplicates, stop-list, Top N) exist for a fast visual check without reading line by line.
- Use Batch Tools for repetitive edits — if you notice after a batch that, say, every image needs duplicates removed, don't open every card — one click in Batch Tools applies it to all of them, fully undoable with Ctrl+Z.
- On Free, register a Puter account instead of staying a guest — a guest session doesn't carry progress/quota between visits, while a registered account has a stable monthly quota.
- Mind each platform's keyword requirements — Dreamstime rejects under 7 keywords, Adobe requires exactly 49 — picking the right platform profile up front (not Universal) saves manual adjustments later.
- Save with EXIF/XMP vs. CSV Export — different scenarios: EXIF/XMP helps when a platform reads metadata straight from the file on upload; CSV is for platforms with a separate bulk-upload template.
- If a batch pauses unexpectedly — check the banner text, it usually means your balance/quota ran out or your session expired, not a generic error. Sign in again or wait for the next quota, then hit "▶ Resume" to continue, or "✕ Cancel & keep results" if you're happy with what's already there.
- Start small on big batches — see "File and Batch Limits" in Getting Started; a few smaller batches beat losing one big batch to a frozen browser tab.
1. Generation Settings
1.1. AI Model
Dropdown of models that analyze video (picture + sound). Picking a model immediately shows its Credits VB price per second of video at your current settings.
1.2. Platform
21 platform-specific profiles (section 3) — video has no universal profiles the way Image does; every platform has its own distinct requirements. Platform sets the prompt and default field limits.
1.3-1.6. Title/Description, Keywords, Top Keywords, Keyword Rules
Same elements and behavior as Image (see the Image tab, sections 1.3-1.6) — field toggles, Include/Exclude/Trademarks, Auto-apply, Auto-remove Duplicates. Platforms with no separate Title field (TikTok, Twitter/X, and others — see the table in section 3) simply don't show a Title toggle, only Description.
2. Custom Platforms (Pro)
Up to 5 saved profiles for video (a separate pool from Image). Your own name + instructions for the AI, fully replacing the preset platform via the "Use Custom Platform" toggle.
3. Platform Profiles
3.1. Premium video stocks
| Platform | Title / Description | Keywords | Notable rules |
|---|---|---|---|
| 🅐 Adobe Stock | Title ≤ 200 (aim 70) · Desc ≤ 200 | Exactly 49 | Order: main subject/action → setting/camera work → general concepts. No brands. |
| 🔴 Shutterstock | No Title field — one Description ≤ 2048 combining headline and description | 50 (min. 7) | The first sentence works as the headline. Editorial format for newsworthy scenes. English only. |
| 📷 Getty / iStock | Title ≤ 200 · Desc ≤ 500 | 50 | Mandatory people-count tag, camera movement if notable (static/panning/tracking/aerial), controlled vocabulary. Getty doesn't publish an exact keyword cap for video — 50 is the top of the practical 25-50 range from contributor guides. |
| 🎞️ Pond5 | Title ≤ 80 · Desc ≤ 2000 | 50 | The first 10 keywords carry more search weight — put the most specific terms first. Editorial date format for newsworthy scenes. English only. |
| 🎛️ Motion Array | Title ≤ 100 · Desc ≤ 500 | 20 | Sells effects/transitions/templates for editors — asset type is required within the first 5 keywords. Exact official limits aren't published (owned by Artlist since 2023-24) — a working estimate. |
3.2. Mid-tier video stocks
| Platform | Title / Description | Keywords | Notable rules |
|---|---|---|---|
| 🟦 Depositphotos | Title ≤ 100 · Desc ≤ 300 | 35 | Short and to the point, no brands. |
| 🟤 Dreamstime | Title ≤ 100 · Desc ≤ 500 (mandatory) | 50 (min. 7) | Description is mandatory — don't leave it thin. The first 7-10 keywords matter most. |
| 🟧 123RF | Title ≤ 100 · Desc ≤ 500 | 25 (min. 7) | Short and to the point, no brands. |
3.3. Free video stocks
| Platform | Title / Description | Keywords | Notable rules |
|---|---|---|---|
| 🟩 Pixabay | No Title · Desc ≤ 500 | 15 | Pixabay's own guide recommends 10-20 tags. AI-generated content must be honestly labeled as such (platform requirement). |
| 🟢 Pexels | No Title · Desc ≤ 200 | 8 | Pexels' own upload form makes title/tags optional — Description is the field that matters most here. |
| ⬜ Unsplash Videos | No Title · Desc ≤ 300 | 5 (for your own reference) | Unsplash auto-tags from the description via its own ML — regular contributors have no manual tag field, Keywords here is just for your own organization. |
3.4. Social media
| Platform | Title / Description | Keywords/tags | Notable rules |
|---|---|---|---|
| ▶️ YouTube | Title ≤ 100 · Desc ≤ 5000 | 10 | The first 3-5 words of the title are critical for search. In 2026 tags carry little ranking weight — don't over-invest in their number. No more than 15 hashtags in the description (or YouTube strips them all). |
| 🎵 TikTok | No Title · Desc ≤ 2200 | 20 | The first 100-150 characters are the in-feed hook. Hashtags are written as plain words, no #. |
| 📸 Instagram Reels | No Title · Desc ≤ 2200 | 8 | The first ~125 characters — the tightest cutoff of any social platform. Technical hashtag cap is 30, but Meta itself recommends 3-5 precise ones over a big list. |
| Title ≤ 100 · Desc ≤ 500 | 8 | Title only applies when uploading as a video asset (Business Suite), not a regular post. Hashtags underperform here versus Instagram/TikTok — 3-5 is usually enough. | |
| 🐦 Twitter/X | No Title · Desc ≤ 280 | 5 | Description is the tweet text itself. In practice 1-2 hashtags work best. |
| No Title · Desc ≤ 3000 | 5 | The first 140-210 characters show on mobile before "see more". Professional tone, 3-5 focused hashtags. |
3.5. Professional platforms
| Platform | Title / Description | Keywords/tags | Notable rules |
|---|---|---|---|
| 🎮 Twitch | Title ≤ 140 · Desc ≤ 300 | 10 | Tags: up to 25 characters each, no spaces (camelCase for multi-word, e.g. "speedRun") — that's Twitch's actual tag format, not just style. Description isn't officially documented for VOD uploads — a working estimate. |
| 🎬 Vimeo | Title ≤ 128 · Desc ≤ 5000 | 20 | Description is written as 1-2 flowing paragraphs (not a keyword list) — unlike social platforms, a thorough description genuinely helps discovery on Vimeo. Can include segment timestamps. |
| ✈️ Telegram | No Title · Desc ≤ 1024 | 10 | The caption limit for a standard (non-Premium) account — deliberately the stricter number, so a caption doesn't get rejected on send. |
| 🎧 Discord | No Title · Desc ≤ 2000 | 8 | Discord has no tag field for regular video messages — Keywords here is just for your own organization. Discord Nitro raises the text limit to 4000. |
4. Upload and Generation
4.1. Uploading files
Drag files into the dropzone or browse via the file picker. Formats: MP4, MPEG, MOV, AVI, FLV, WebM, WMV, 3GP. Up to 50 files per batch, up to 200 MB per file, duration from 5 seconds to 60 minutes — shorter or longer files are rejected before any credits are charged (see also "File and Batch Limits" in Getting Started — a technical ceiling, not a guarantee for every computer).
4.2. Processing — 2 real phases
Unlike Image (one request with the file inside), video goes through two steps: first the file uploads to the AI provider's side ("Uploading..."), then it waits to be processed ("Google is processing the file..."), and only then does metadata generation itself run ("Generating metadata..."). The progress bar on each card reflects the real phase, not an animation — which is why video takes noticeably longer than image.
Error classification and batch-pause behavior on a blocking error work the same as Image (see the Image tab, section 5.2).
5. Result Card Interface
Same elements as Image (see the Image tab, section 6): Title/Description with a live counter, the keyword chip editor, the card tools menu, global Undo/Redo history — minus categories, which video doesn't have.
💾 Save with XMP — saves the file (MP4/MOV) with metadata embedded directly, no extra credits. Whether the metadata already in the file (title, author and comment written by your camera or editor) is kept or replaced is controlled by Existing metadata in Settings.
6. 🧰 Batch Tools
Same mechanism as Image (section 7 there) — applies a Keyword Rules action to every completed card at once, undoable with a single Ctrl+Z.
7. 📄 CSV Export
Same module and presets as Image (see the Image tab, section 8 — Universal, Adobe, Shutterstock, Magnific, Dreamstime, Vecteezy, Getty/iStock, Pond5, plus the custom column builder).
8. Bulk Export
Download All as ZIP and Save All to Folder work exactly as they do on Image (section 9 there).
9. Generation History (/pro-history)
Off by default, turned on in Settings — same as Image (section 10 there).
10. Usage Tips & Recommendations
- Pick a platform profile up front, not Universal — video has no universal profiles, so the right platform choice saves manual limit adjustments later.
- Mind platforms with no separate Title (Shutterstock, Pixabay, Pexels, Unsplash, TikTok, Instagram, Twitter/X, LinkedIn, Telegram, Discord) — everything lives in Description there, so make sure the opening characters (the ones shown before any cutoff) carry the main point.
- Vimeo — write a full description: unlike social platforms, it genuinely helps discovery there, not just fills space.
- YouTube — don't over-invest in tags: they carry little ranking weight in 2026; the title and the video's first seconds matter far more.
- Set up Trademarks and the stop-list once — same as Image, saves per-card edits.
- Save with XMP vs. CSV Export: XMP when a platform reads metadata straight from the file; CSV for platforms with a separate bulk-upload template.
- If a batch pauses unexpectedly: check the banner — usually your balance ran out or there's a temporary issue on our side, not a generic error. "▶ Resume" continues from where it stopped.
- Start small on big batches: video is heavier on the browser than image — see "File and Batch Limits" in Getting Started.
1. Generation Settings
1.1. AI Model
Dropdown of models that analyze audio. Picking a model immediately shows its Credits VB price per file at your current settings.
1.2. Platform — three different field groups
9 profiles split into three groups, each with a fundamentally different result field set (details in section 3):
- Sound & music (regular Title/Description/Keywords): Pond5 Sounds, Freesound.org, SoundCloud, YouTube Podcasts, Telegram.
- Podcasts (Episode Title/Episode Description — about a specific episode, not the show as a whole): Spotify for Creators, Apple Podcasts, Podbean.
- Audiobooks (Genres instead of Keywords, up to 3): Google Play Books.
1.3. Title/Description/Keywords — ranges, not fixed numbers
Unlike Image/Video, where a platform sets one default limit, every Audio platform defines an allowed range for each field (Keywords: 10-50 on Pond5 Sounds, but 0-20 on YouTube Podcasts, for example) — you choose the exact value within that range. Field on/off toggles work the same as Image/Video.
1.4. Keyword Rules, Top Keywords
Same elements as Image/Video (see the Image tab, section 1.6) — Include/Exclude/Trademarks, Auto-apply, Auto-remove Duplicates, Top Keywords Count ranking. For the Audiobooks group (Genres instead of Keywords), these rules don't apply — genres aren't filtered like keywords.
1.5. BPM Detection
An optional toggle — detects a track's tempo right in your browser (nothing sent to a server) and feeds it into the prompt. Useful for music platforms (Pond5 Sounds, Freesound, SoundCloud). Not every track allows a reliable tempo read — if it fails, the field just stays empty, with no effect on the rest of the result.
1.6. Show Profiles (Podcasts group)
A saved show profile — name, language, hosts. Set up once in Settings, applied to every new episode instead of re-entering the same details.
1.7. Author Name (Google Play Books only)
The audiobook's author name — fed into the prompt for the Audiobooks group.
2. Custom Platforms (Pro)
Up to 5 saved profiles for audio — your own name + instructions for the AI + which field set to use (regular/episode/genre). Fully replaces the preset platform via "Use Custom Platform".
3. Platform Profiles
3.1. Sound & music
| Platform | Title | Description | Keywords/Tags |
|---|---|---|---|
| 🎧 Pond5 Sounds | 20-80 chars | 50-500 chars | 10-50, top up to 20 |
| 🎵 Freesound.org | 20-300 chars | 50-2000 chars | 5-50 tags, top up to 20 |
| ☁️ SoundCloud | 10-100 chars | 50-4000 chars | 5-40 tags, top up to 15 |
| ▶️ YouTube Podcasts | 20-100 chars | 200-5000 chars | 0-20 tags (carry little ranking weight in 2026) |
| ✈️ Telegram | — (no separate title) | 100-1024 chars (caption) | 0-20 hashtags, top up to 10 |
Notable rules: Freesound requires disclosing AI-generated audio (a "GenAI" tag + model name in the description when the Settings toggle is on), plus a separate explicit-content flag. SoundCloud adds Artist, Record Label, Buy Link, and Album fields in Settings (one value per CSV export, not fed into the prompt).
3.2. Podcasts (Episode Title / Episode Description)
| Platform | Episode Title | Episode Description |
|---|---|---|
| 🎧 Spotify for Creators | 30-300 chars | 200-8000 chars |
| 🎙️ Apple Podcasts | 30-300 chars | 200-10000 chars |
| 🎛️ Podbean | 30-300 chars | 200-8000 chars (tags: 5-500, depending on your Podbean plan — 20 on Free, up to 500 on paid tiers) |
The result card for this group also shows Category/Subcategory (Apple/Spotify/Podbean podcast taxonomy, picked right on the card) and a per-episode Explicit toggle (defaults from the show profile, editable per episode).
3.3. Audiobooks (Genres instead of Keywords)
| Platform | Title | Description | Genres |
|---|---|---|---|
| 📚 Google Play Books | 20-300 chars | 200-3000 chars | 1-3, BISAC-style categories |
4. Upload and Generation
4.1. Uploading files
Formats: WAV, MP3, AIFF, AAC, OGG, FLAC. Up to 50 files per batch, up to 200 MB per file, duration up to 9.5 hours per file — room for anything from a short sound effect to a full podcast episode or audiobook (see also "File and Batch Limits" in Getting Started).
4.2. Processing
The same two real phases as Video (upload → processing on the AI provider's side → generation), plus an optional "Detecting BPM..." phase when BPM detection is on. Error classification and batch-pause behavior work the same as Image/Video.
5. Result Card Interface
Which fields appear depends on the platform's group (section 3): Title/Description/Keywords, or Episode Title/Episode Description (+ Category/Subcategory/Explicit), or Title/Description/Genres. Otherwise, the same chip editor, card tools menu, and global Undo/Redo history as Image/Video.
💾 Save with Metadata (no extra credits) — MP3 gets a full set of ID3 tags (plus Artist/Album/Publisher from Settings and BPM/Duration when known), WAV/AIFF/FLAC get XMP/Vorbis Comment depending on the format.
6. 🧰 Batch Tools
Same mechanism as Image/Video — applies a Keyword Rules action to every completed card at once (except the Audiobooks group, which has no keywords to bulk-edit).
7. 📄 CSV Export
Its own field catalog (not shared with Image/Video — audio platforms take bulk uploads through different formats):
- 🌐 Universal (basic) — Filename, Title, Description, Keywords.
- 🎧 Pond5 Sounds — OriginalFilename, Title, Keywords (the only platform with an officially confirmed bulk-CSV format — the same importer Pond5 Footage/Photos uses).
- 🎵 Freesound — Filename, Title, Description, Keywords, License, AI Generated, AI Model (the platform doesn't publish an official format; this is a working set covering its required fields).
- 🛠 Custom — build your own from the catalog: beyond the standard fields there's Duration, BPM, Category/Subcategory (podcasts), License/Artist/Record Label/Buy Link/Album (SoundCloud/Freesound), Author (Google Play Books), and ISRC/ISBN (always empty — filled in by hand per track/book, since each one is unique).
8. Bulk Export
Download All as ZIP and Save All to Folder work exactly as they do on Image/Video.
9. Generation History (/pro-history)
Off by default, turned on in Settings.
10. Usage Tips & Recommendations
- Pick the right platform group first: switching between Flat → Episode → Genre completely changes the field set — decide on the content type before starting a batch, so you don't spend credits regenerating.
- Set up a Show Profile once per podcast: saves time on every new episode.
- BPM detection is for music platforms only: turning it on for podcasts/audiobooks does nothing, since that field isn't used there.
- SoundCloud/Freesound — fill in Settings once: Artist/Album/Label/License apply to every row in the CSV automatically, no need to type them per card.
- Google Play Books — set Author Name in Settings before an audiobook batch.
- ISRC/ISBN — always manual: these are unique per track/book, so the system deliberately leaves them blank in the CSV.
- If a batch pauses unexpectedly: same as Image/Video — check the banner, "▶ Resume" continues from where it stopped.
- Start small on batches of long files: several multi-hour audiobooks in one batch is especially demanding on the browser — see "File and Batch Limits" in Getting Started.