Writing and testing
Şükrü Çelik is an independent software developer with hands-on training in computer engineering and modern full-stack web architecture, focusing on cloud systems and AI (LLM) integrations. At Polytested, Şükrü tests AI tools on technical criteria such as data privacy, integration and real-world performance rather than marketing claims.
What keyword search volume counts, how Google, Semrush and Ahrefs get their numbers, why they differ, and how to check volume per country for free, with our own Semrush volumes next to our Search Console impressions.
What striking distance keywords are, why position 11 isn't always page two, and how to find, pick and improve them in Search Console, with our own two weeks of data and one title change.
How to check where your site ranks on Google for a keyword: free in Search Console, with rank trackers, and why your own search misleads. With our own data: one page, positions from 8 to 49 for different searches.
Converting ASS to SRT keeps the text, timing and italics, but not fonts, positions, karaoke or drawings. Our own sample file shows what goes, and how FFmpeg, Subtitle Edit and Aegisub each decide.
A TXT transcript has no timing, so converting it to SRT means adding timecodes. Three ways to do it, a short script, and what our sample showed in four languages when checked against reading-speed limits.
Sub keeps the original voices with subtitles, dub replaces them. Why the two never say the same thing, what studies found, who needs which, and how to switch on Netflix, Prime Video, Disney+, YouTube and Crunchyroll.
One invented conversation written six ways, ready to copy: verbatim, clean verbatim, edited, time-coded, as a research interview and as a video transcript, with what each style keeps, drops and is good for.
How to turn a recording or MP3 into text with Whisper on your own computer, Word, an iPhone or upload sites, and how accurate and fast eight free speech engines were in our test with real audiobook speech.
How a small business gets found nearby: setting up and verifying a Google Business Profile, reviews within Google's and the FTC's rules, the website basics, and what our survey of 63 local business websites showed.
YouTube's own transcript is free and often all you need. How to open, search and copy it in the three panel versions we found, fix glued timestamps, and what generator sites and AI summaries add.
What "low" and "0" mean in Semrush, Ahrefs, SISTRIX and Google, when a small search deserves a page or just a section, and which searches our new site actually appeared for in its first days in Search Console.
n8n is a workflow automation platform from Berlin. How nodes, triggers and executions work, what the license allows, where your data lives and how secure it is, with an invoice example we tested and data from 13,300 templates.
Have any text read aloud for free in Edge, Chrome or Firefox, with Windows, Word or Acrobat Reader, on Android, iPhone or Mac. With the English voices we found in Edge and Chrome, and which voices send your text to a server.
White hat SEO follows the search engines' rules. Google's 16 spam policies in one table, ten everyday cases, what a violation costs, why Europe and the US now differ, and how we apply the rules on our own site.
SEO helps search engines find, understand and show your pages in their free results. We explain how it works in plain English, with what Google's crawler did when we opened our own site in October 2026.
How long should an SEO title be, and how do you structure it? What Google's documentation says, why character counts mislead, and our own titles measured in pixels, with weak and better examples.
How to set up a site for several countries and languages: URL structure, hreflang, no automatic redirects, keywords per market, and what changes in the US, Canada, the UK and Europe. With our own four-language site as the example.
Collect old URLs, map 301 or 308 redirects, unlock staging, test with curl and read Search Console afterward. With the real server responses and Search Console data from our own change of 35 URLs in October 2026.
Why a translated keyword is rarely the one people search, which translation method fits search, what Google says about machine translation, and the workflow we use for four languages, with our own test data.
How to find keywords in five steps: collect ideas for free, check volume and difficulty, look at who ranks and give each topic one page. Shown with our own US plan of 95 topics, and what ChatGPT can and can't do.
How to turn YouTube subtitles on or off on a computer, phone or TV, keep them off for every video, change the language and the caption style, and why captions keep switching on by themselves. Checked in October 2026.
A .vtt file holds captions in the WebVTT format: a WEBVTT header, timed cues and text. Our sample file line by line, VTT vs. SRT, how to open and convert it, and what made captions vanish in our Chrome test.
Intelligent automation means AI steps inside otherwise rule-based workflows. What it is, how it differs from RPA, hyperautomation and AI agents, what it costs, and what we found in all 13,300 templates of n8n's public library.
We pasted the same SRT file into Google Translate, DeepL and Bing, in German, French and Spanish: which kept it intact, where the free limits cut long files, and why every translation still needed editing for length and reading speed.
Where to put your keyword, place by place, with what Google and Bing actually document and which popular rules have no source at all. Plus an audit of where the target keyword sits in our own 21 guides.
What a keyword is, what search volume, difficulty and intent really tell you, and how to pick keywords a new site can win, shown with the 305 target keywords we chose for our own guides in four markets.
Which hreflang value to use for Spanish pages, why Google doesn't support es-419, four setups that work, and what 18 big multilingual sites actually declare, from a plain es to codes that don't exist.
A canonical tag tells search engines which URL is the main version of a page. What it looks like, how Google chooses, when to use a redirect or hreflang instead, and how our own site answers 11 versions of its URLs.
LRC files only say when each line starts; SRT also needs an end. How to convert with Subtitle Edit, FFmpeg or an online tool, what converters guess, and why YouTube takes LRC as it is. Shown with our own lyric file.
Turn subtitles into plain text you can read, search or summarize: with Subtitle Edit, four find-and-replace steps in any text editor, or an online converter. Tested on our own SRT file.
Dubbing replaces a video's voices with new ones, usually in another language. How it differs from voice-over and subtitles, how studios and AI tools make dubs, where audiences expect them, and what we saw in YouTube's audio track menu.
TikTok shows captions but has no transcript or export. What we found in TikTok's web player, how to turn captions and translations on, the clean route for your own videos, and what transcript generators actually do.
Lovable apps can rank, but what a crawler gets depends on when the app was made. We fetched nine public lovable.app sites: what came back, what Lovable handles for you, and the checks that are still yours.
The steps for every major browser and phone, where the page's text goes when you translate it, and what broke when we ran our own German page through Google Translate.
Consent and audio first, then a style: full verbatim, clean verbatim or edited, shown on one passage. A notation key to copy, realistic time, software, interviews in other languages and the final check.
TTS means text-to-speech: software that turns written text into speech. How it works, where you meet it, the voices we found in Chrome and Windows on our own computer, and what goes wrong in German, French and Spanish.
Translate any YouTube video's captions with Auto-translate, switch to dubbed audio where it exists, and, as a creator, add translated subtitles, titles and your own dubs. Checked in October 2026 on a real video.
Download the caption file of your own videos in YouTube Studio, copy the transcript of any video, and turn a transcript into a valid SRT file, plus what YouTube's terms say about subtitle downloader sites.
Who really competes with you in search, how to pull their keywords and gaps in Semrush or Ahrefs, the free alternatives, and how far to trust the numbers, with our own data from 200 searches in the US, Germany, France and Spain.
Subtitles carry the dialogue for viewers who don't understand the language; closed captions add the sounds for viewers who can't hear them. Shown with our own sample files in four languages, plus SDH, open captions, labels and the rules.
What meeting minutes are and what to put in them, with our template, a filled-in example, the rules for formal minutes, the words used in German, French and Spanish, and what AI note-takers can draft for you.
Get the original audio back on one video in three clicks, stop most dubs and translated titles with Preferred languages, and switch automatic dubbing off as a creator. Steps checked in October 2026, with what we saw on a dubbed video.
How we researched keywords for the US, Germany, France and Spain: the same topic is a different query, intent and difficulty in each market. Our method and our own Semrush data, with the checks we wouldn't skip again.
Integromat was renamed Make in February 2022 and switched off in September 2023. What happened to accounts and scenarios, where the old links lead today, and what has changed in Make since, mapped to n8n and Zapier.
How hreflang, canonical and x-default work together on a multilingual site, shown on our own four-language setup: the tags in the head, the sitemap, untranslated pages and the mistakes that make Google ignore your annotations.
A time-coded transcript ties every passage to a moment in the recording. The timestamp styles side by side, how they differ from SMPTE timecode and caption files, and five ways to make one, from free and offline to fully automatic.
An SRT file is a plain-text subtitle file: numbered blocks with a start time, an end time and a line of text. We built one in English, German, French and Spanish to show what a wrong encoding does, and how to create, translate and upload one that works.