Text-only Shorts on YouTube: when scrolling text counts as value

Text-only Shorts are not forbidden, but scrolling text with minimal or no narrative is on the not-allowed list for inauthentic content, next to image slideshows and templated storylines (read 2026-10-03). The qualifier is the point: text that teaches something, argues something or tells a story is the video. Text that restates a headline over a background loop is the template.

This is a live question after October 1, because the update is reported to cut reach for re-uploaded clips and to discount template-based bulk changes (Search Engine Journal, read 2026-10-03). A text Short is easy to batch, which is why it gets read as a template unless each one has content that could not be swapped.

What separates a text Short from a template

Count the words that only you could write. If that number is close to zero, add voice or rewrite.

Timing text to a clip with caption cues

Video captions accepts cues (or segments) with text and start and end times in seconds, which skips speech-to-text and burns exactly that copy at those times. The docs describe this as the path for silent clips, since speech-to-captions on a silent clip finds nothing to transcribe. A caption job costs $0.20 per clip up to 60 seconds.

The helper below turns a script into cues at a reading speed you choose, one line per card, with a short gap between. It keeps each card readable, which is where many text Shorts fail. Check the exact request fields against the captions doc before use; the video must be a fetchable HTTPS URL.

import os, requests

def cues_from_lines(lines, wps=2.6, gap=0.25, min_secs=1.4):
    cues, t = [], 0.0
    for text in lines:
        secs = max(min_secs, len(text.split()) / wps)
        cues.append({"text": text, "start": round(t, 2),
                     "end": round(t + secs, 2)})
        t += secs + gap
    return cues

lines = ["Most budget apps fail at one thing.",
         "They ask you to log, not to decide.",
         "Try one weekly decision instead."]
cues = cues_from_lines(lines)
print(cues[-1]["end"], "seconds of text")

body = {"video_url": "https://media.sume.com/example/silent-loop.mp4",
        "style": "slam", "cues": cues}
key = os.environ.get("SUME_API_KEY")
if key:
    r = requests.post("https://api.sume.com/v1/video-captions", json=body,
                      headers={"Authorization": f"Bearer {key}",
                               "Idempotency-Key": "text-short-001"})
    print(r.status_code)

Making the video behind the text

The base clip needs a length that matches the last cue. Video trim cuts a range from one hosted clip with a duration from 0.2 to 900 seconds, so a longer loop of your own footage can be trimmed to fit. Poll the job through GET /v1/jobs/:id/status, as in the jobs docs, then read the result.

If the Short's value is in what you say, record it. A voice over a text card is the minimum that separates 'I read this to you' from a template.

What this cannot promise

Sume cannot tell you how YouTube classifies a text Short and does not guarantee reach. The captions tool burns the words you give it and says nothing about whether they are worth reading. Use the checklist: a claim, evidence you chose, a point of view, a different angle each time.

Reading speed and length

A card that is gone before it is read is also noise. The helper above uses 2.6 words per second and a floor of 1.4 seconds per card; those are my defaults, not a Sume or YouTube rule, so tune them to your audience. Shorts can be up to 3 minutes in vertical format (read 2026-10-03), but a text-only Short rarely earns that length.

If the message needs more than about 12 cards, consider recording it as speech instead and letting the captions follow.

Three text Shorts that carry value

A debug story told in six cards, each a step you took and what it showed, ending with the cause. A comparison with a table of numbers you measured, shown one row at a time, ending on your pick. A contrarian claim, one card of evidence per objection.

Each has a spine that you wrote, which is the difference from a headline reel. None depends on the picture behind the text, so the background can be your own simple loop.

If you want sound, a music bed from Music 1.0 is a text prompt with no duration parameter, so plan to trim or loop it to the cards. Whether music changes how YouTube sees the Short is a separate question covered in this post.