HappySRT
Open-source web workspace that turns audio and video into timestamped transcripts, multilingual subtitle tracks and bullet-point summaries. Runs the three steps separately or chained, exports SRT or plain text, and meters usage through media tokens.
What is HappySRT?
HappySRT is an open-source web application that turns recorded audio and video into three kinds of text output: a timestamped transcript ready for the SRT subtitle format, translated subtitle tracks, and a condensed summary made of highlights and bullet notes. The whole product sits behind a single page; opening the site loads a browser workspace that checks the session, starts an in-browser FFmpeg engine and syncs previous work.
Work is organised into threads, one per project, such as a podcast episode, a client or a course module. Each thread keeps its own uploads, outputs and exports, so results stay grouped instead of piling into one undifferentiated list. Inside a thread the user uploads an audio or video file, or pastes a media URL, then runs a single action or chains all three.
Transcription can be routed to one of four speech recognition engines, each with its own cost per minute of media: Deepgram Nova-3 at 24 tokens per minute, Deepgram Whisper at 30, UpliftAI Scribe at 18 and UpliftAI Scribe Mini at 12. Translation and summarisation are handled by GPT-4o mini, with automatic source-language detection and forty target languages ranging from French and Spanish to Arabic, Hindi, Japanese and Filipino. Summaries default to ten bullet points.
Consumption is metered in media tokens, presented in the interface as processing credits; the balance sits in the sidebar and tokens show as in use while a job runs. Results appear as Transcription, Translation and Summary tabs, and leave the tool through Download SRT, Copy or Save.
Visitors can try the app as a guest, while signing in with Google or with an email and password is what makes threads persist and sync across devices. The application installs as a progressive web app, though no native mobile app exists. Its source code is published on GitHub and can be self-hosted, which requires supplying an Appwrite project, Stripe keys and AI provider credentials.
What the site does not publish matters as much as what it does: no terms of service, no privacy policy, no legal notice, no company name, no postal address and no contact email.
What it does
- Transcribe an audio or video file into timestamped, SRT-ready text
- Translate a transcript into subtitle tracks across 40 target languages
- Summarise a recording into highlights and bullet notes
- Chain the three actions in a single run: transcribe, translate, summarise
- Export results as SRT or plain text through Download, Copy or Save
- Organise projects into separate threads, each with its own uploads and outputs
- Pick the speech recognition engine and let the source language be detected automatically
When to use HappySRT / When not to
A quick filter to help you decide if HappySRT is the right fit.
When to use HappySRT
- Video creators who need clean, timestamped SRT files for their own footage
- Podcasters wanting an episode transcript and a bullet summary in one pass
- Journalists and documentalists transcribing interviews and archive recordings
- Translators and localisation specialists who want a first subtitle track to post-edit
- Teachers, students and researchers turning recorded lectures or interviews into searchable notes
When not to use HappySRT
- Organisations that need a contract: no terms of service, no privacy policy and no DPA are published
- Regulated sectors such as healthcare, law and finance, where no certification or confidentiality commitment is offered
- Teams building on an API: none is documented and no public API route exists
- Buyers who need an identifiable vendor, since no company name, address or contact email appears anywhere
- Anyone needing human support or long-term archiving, with no support channel published and free-tier retention capped at 14 days
How to use HappySRT
A typical end-to-end flow, from setup to results.
- Open happysrt.com and let the workspace load in the browser, including the session check and the FFmpeg engine
- Explore in guest mode, or sign in with Google or an email and password to keep your history
- Click + New thread to create a dedicated workspace for the project
- Upload an audio or video file, or paste a media URL
- Choose the speech recognition engine among the four offered, each with a different token cost per minute
- Set the source language or leave automatic detection on, then pick one or more target languages
- Run a single action, or launch the full chain: transcribe, then translate, then summarise
- Watch the media token balance in the sidebar while the job is processing
- Read the results in the Transcription, Translation and Summary tabs of the thread
- Export with Download SRT, Copy or Save, and open Upgrade from the sidebar if the token balance runs low
Pros & Cons
Pros
- Three complementary treatments in one tool, chainable in a single run
- Permanent free plan with 100 media tokens a month and no card required
- Very low paid entry point at 5 USD per month
- Public source code, auditable and self-hostable with your own provider keys
- Explicit choice of speech recognition engine, with its token cost shown per minute
- Forty translation target languages plus automatic source detection
- Unusually detailed and fully quantified plan grid, down to storage and retention limits
Cons
- No legal documents at all: no terms of service, no privacy policy, no legal notice
- Completely anonymous vendor, with no company name, postal address or email
- No published support channel of any kind
- No public API and no API documentation
- Short retention on the free plan, where data may be dropped after 14 days
- Open source is claimed but no licence is actually granted, the repository still carrying the template line
- Pricing is visible only inside the application, never on a public page, and the project shows almost no community traction
Pricing & Plans
A permanent free plan is available at no cost, providing 100 media tokens per month, two threads and 14 days of retention. The lowest paid entry point is the Hobby plan at 5.00 USD per month. Billing is monthly only, with no annual option offered, and payments are processed through Stripe.
- 100 media tokens per month
- 2 threads
- 100 MB maximum file
- 512 MB monthly upload
- 2 GB active storage
- 14-day retention
- 50 saved items
- no add-ons
- 1
- 000 media tokens per month
- 20 threads
- 500 MB maximum file
- 10 GB monthly upload
- 50 GB active storage
- 90-day retention
- 2
- 10
- 000 media tokens per month
- 200 threads
- 2 GB maximum file
- 50 GB monthly upload
- 1 TB active storage
- 180-day retention
- 20
- 500
- 000 media tokens per month
- 2
- 000 threads
- 10 GB maximum file
- 1 TB monthly upload
- 5 TB active storage
- 365-day retention
- adds 3
- 000 media tokens per month
- 25 GB of monthly upload and 100 GB of active storage
- available on paid plans only
- 2
- 000 media tokens for 10 USD as a one-off purchase
- roughly 0.005 USD per token
Data, GDPR & hosting
A consolidated view of how HappySRT handles your data.
GDPR overview
There is no mention of the GDPR anywhere: not on the site, not in the application bundle, not in the public repository. This is an absence, not an omission in our review. No data controller is identified, since no company name or postal address is published. No data protection officer is named and no representative is designated under Article 27. No contact email exists, so there is no documented route for exercising access, rectification or erasure rights. No legal basis, no contractual retention period and no subprocessor list are disclosed, although the service demonstrably routes user media through several third parties for speech recognition, translation, storage, payment and analytics. For a user in the European Union, none of the information the GDPR expects to find is available.
Who owns the data?
No answer can be given from the vendor's own documents, because there are none. HappySRT publishes no terms of service, no privacy policy and no legal notice; the site's route table contains only the application, its billing callbacks and a password reset page. Nothing therefore states who owns an uploaded recording, a generated transcript or a translated subtitle track, nor what the operator may do with them. The only statement resembling a commitment is an in-app note saying the app is creator-friendly and that users control what they upload and export, which carries no contractual weight. Storage runs through Appwrite, and the public repository advises treating uploaded media and transcripts as sensitive data.
Reuse rights
In practice users can reuse their outputs freely, and the product is built for it: results are exported through Download SRT, Copy and Save, in SRT or plain text, with no watermark, quota on reuse or approval step. That freedom is a consequence of the design rather than a granted right, since no terms of service define or limit any licence over the outputs. The same gap affects the software itself: the project is presented as open source, yet the repository's licence section is still the untouched template line inviting the author to add one, so no licence is actually granted over the code either.
Data retention & training
Hosting summary
The vendor declares no hosting jurisdiction whatsoever: no country, no region, no data centre, and no commitment about where uploaded media, transcripts or account data reside. What can be observed is purely technical and carries no contractual value. The domain is served behind Cloudflare, resolving to an anycast address, and the application bundle points its backend at Appwrite Cloud for authentication, database and storage. Those are infrastructure facts inferred from the code, not statements made by the operator, and the hosting regions of Cloudflare or Appwrite belong to those providers rather than to HappySRT. The public repository mentions that S3 or B2 storage can be configured, but that applies to whoever deploys their own instance, not to the hosted service. Anyone with a data residency requirement has nothing here to rely on.
Things to keep in mind
Risks and trade-offs to weigh before adopting HappySRT.
- No contractual framework at all: nothing governs what happens to an uploaded recording once it leaves your machine
- No identifiable vendor means no recourse and no counterparty if something goes wrong or data leaks
- No published training policy, so it is impossible to know whether your content feeds any model
- An undisclosed subprocessor chain, while media demonstrably passes through third parties for recognition, translation, storage and payment
- Retention is a plan quota, and on the free tier work can disappear after 14 days if it is not saved elsewhere
- Trusting automatic output unchecked: an unreviewed transcript or translation can distort names, figures and meaning, and published subtitles carry that error to the audience
- Uploading copyrighted material or private conversations recorded without the consent of the people involved
Setup & Integrations
Technical difficulty
For ordinary use there is nothing to install and no technical difficulty: HappySRT is a web application that can be tried in guest mode, with sign-in available through Google in one click or through an email and password. It can optionally be installed as a progressive web app from the browser. The only requirement is a recent browser, since FFmpeg runs client-side. Self-hosting is a different matter and clearly aimed at developers: it means supplying your own Appwrite project for authentication, database and storage, your own Stripe keys, your own AI provider credentials and a local environment file.
Deployment
Integrations
Supported languages
Behind HappySRT
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does HappySRT actually do?
Is there a free plan, and what does it include?
How much does the cheapest paid plan cost?
How is usage counted?
Which languages and engines are supported?
Do I need an account, and is there a mobile app?
Does HappySRT publish terms of service or a privacy policy?
How long is my data kept?
Can I self-host HappySRT, and is there an API?
Should you pick HappySRT?
HappySRT is technically credible and inexpensive on its core job. Timestamped transcription, multilingual subtitle tracks and bullet summaries sit in one workspace, the three steps chain together in a single run, and the plan grid is unusually precise for a product of this size, quantifying tokens, thread counts, file sizes, storage and retention rather than hiding behind vague tiers. Letting the user pick the speech recognition engine, with its token cost per minute stated openly, is a genuine mark of transparency. The permanent free plan and the 5 USD entry point make it easy to test on real footage, and publishing the source code gives technically minded teams a way out if they would rather run the whole thing themselves.
The counterweight is severe, and it is a matter of governance rather than features. The vendor is entirely anonymous: no company name, no postal address, no email, no support channel. There are no terms of service and no privacy policy, so nothing states who owns an uploaded recording, how long it really survives, or whether it feeds anything else. The GDPR is never mentioned, even though media is demonstrably routed through several third-party providers that are never disclosed. The open-source claim is weakened by a repository that still carries an unfilled licence placeholder, and the project shows almost no community traction.
The sensible reading is a split one. For personal or low-stakes material, where losing a file or having it processed by an unnamed party carries no real consequence, HappySRT is a capable and cheap tool. For client work, interviews with identifiable people, or anything touching a regulated sector, the absence of a contract and of an identifiable counterparty is disqualifying, however good the output looks.
- Choosing a selection results in a full page refresh.
- Opens in a new window.