🎙️ Kick VOD Transcript Scraper — faster-whisper

Turn any Kick.com VOD into a searchable transcript. Kick ships no captions, so this actor pulls the VOD audio and transcribes it with faster-whisper (Whisper large-v3-turbo) — giving you clean text plus SRT/VTT subtitles and timestamped segments, by channel or VOD URL. Run it on a schedule to auto-transcribe new streams as they go up.

No login, no browser automation. Metadata comes from Kick's public API; the transcript is generated on Apify's own compute — no external API key needed.

---

What does it do?

Kick exposes VODs through a public JSON API, but none of them have a caption/subtitle track — so the only way to get a transcript is speech-to-text. This actor does exactly that:

1. 🔎 Finds VODs — newest N per channel, or specific VOD URLs.
2. 🎧 Extracts the audio — pulls the lowest-bandwidth HLS variant (audio intact) and converts it to 16 kHz mono with ffmpeg. A maxMinutesPerVod cap keeps long streams affordable.
3. 📝 Transcribes with faster-whisper — model large-v3-turbo by default (the speed/accuracy sweet spot), with large-v3 and distil-large-v3 options.
4. 📤 Returns VOD metadata + transcript as plain text, SRT, VTT, and/or timestamped segments.
5. 🔔 Monitors — schedule it to transcribe only newly-published VODs and alert via Slack/email/webhook.

---

What data does it extract?

Per VOD:

• 🆔 vod_uuid, 📺 channel, 🏷️ title, 🎮 category
• 🌐 language (detected), ⏱️ duration_ms, audio_minutes (transcribed), 👁️ views, 📅 created_at
• 🔗 vod_url, 🖼️ thumbnail
• 📝 transcript (plain text) — always included
• 🎬 srt, vtt — subtitle formats (optional)
• ⏲️ segments[] — {start, end, text} timestamped lines (optional)
• ✨ is_new (monitor mode), 🕒 scraped_at

---

Who is it for?

• 🔎 Researchers & journalists searching what streamers actually said, by keyword.
• ✂️ Clippers & editors finding moments fast with timestamped segments + SRT/VTT.
• 🛡️ Trust & safety / brand monitoring scanning streams for terms.
• 🤖 AI / dataset builders assembling speech datasets and feeding RAG pipelines.

---

How to use it (step by step)

1. Click Try for free.
2. Enter one or more channel slugs (e.g. xqc) — or paste VOD URLs.
3. Set Max minutes per VOD (default 60) — Kick streams are long; this bounds cost.
4. Pick a model (large-v3-turbo is recommended) and formats (txt / srt / vtt / segments).
5. Click Start, then open the Dataset tab to view/export.
6. (Optional) set monitorMode + a Schedule + an alert channel to auto-transcribe new VODs.

Quick start

{ "channelSlugs": ["xqc"], "maxVods": 1, "maxMinutesPerVod": 30 }

New-VOD monitor example

{
  "channelSlugs": ["trainwreckstv"],
  "monitorMode": true,
  "maxMinutesPerVod": 120,
  "slackWebhookUrl": "https://hooks.slack.com/services/…"
}

---

Input

Field	What it does
channelSlugs	Kick channels to pull the newest VODs from
vodUrls	specific VOD URLs / UUIDs to transcribe
maxVods	newest VODs per channel
sinceDate	only VODs on/after this date
maxMinutesPerVod	cost guard — cap audio pulled per VOD (0 = whole VOD)
model	large-v3-turbo · large-v3 · distil-large-v3
language / translateToEnglish	spoken language (or auto) / output English translation
transcriptFormats	txt · srt · vtt · segments
includeMetadataOnly	skip ASR, return metadata only (fast/cheap)
monitorMode, alertOnNewVod	recurring new-VOD watcher + alerts
webhookUrl, slackWebhookUrl, emailRecipients	alert channels

---

Output

Each VOD is one dataset record (fields above). Export to JSON, CSV, Excel, HTML, or RSS, or fetch via the Apify API. The transcript is plain text; SRT/VTT are ready to drop into a video editor.

---

How much does it cost?

Pay-per-event — you pay for what you transcribe, no subscription:

Event	What it covers	Suggested price
lot-scraped	each VOD returned	~$0.01 / VOD
audio-minute-transcribed	each minute of audio transcribed	~$0.012 / minute
monitor-run-completed	each scheduled watch run	~$0.05 / run
new-lot-detected	each newly-published VOD	~$0.02 / VOD
alert-delivered	each Slack/email/webhook push	~$0.005 / alert

(Final per-event prices are set on the actor's pricing page. Use maxMinutesPerVod to control spend on long streams.)

---

How does transcription work?

It uses faster-whisper (the CTranslate2 build of OpenAI's Whisper), running on Apify's compute — no API key, nothing leaves Apify. Default model is large-v3-turbo (as accurate as Whisper large-v2 but ~6× faster); distil-large-v3 is cheaper/faster, large-v3 is maximum accuracy. A VAD filter skips long silences so you're not billed for dead air.

---

Is it legal to transcribe Kick VODs?

This actor reads publicly available VOD data via Kick's own public endpoints and transcribes the audio. Transcripts of public broadcasts are widely used for research, accessibility, and indexing, but the underlying streams may be covered by copyright and by Kick's Terms of Service. You are responsible for your use — review Kick's Terms of Service and the applicable rights before redistributing transcripts or audio.

---

FAQ

Why ASR instead of captions? Kick doesn't provide captions/subtitles for VODs, so there's no caption track to extract — a transcript can only be produced by speech-to-text.

Kick streams are 10+ hours — won't that be expensive? Set maxMinutesPerVod (default 60) to bound how much audio is transcribed per VOD. Billing is per transcribed minute, so you control the spend.

Which languages? Whisper supports ~99 languages and auto-detects by default; set language to force one, or translateToEnglish for an English translation.

Can I get subtitles? Yes — add srt and/or vtt to transcriptFormats for ready-to-use subtitle files, or segments for {start,end,text} timestamps.

Can I auto-transcribe new streams? Yes. Turn on monitorMode, attach an Apify Schedule, and add a Slack/webhook/email channel — each run transcribes only VODs new since the last run.

---

Feedback

Want speaker labels, word-level timestamps, or another platform? Open an issue on the actor.