changelog

the previous shrp.app was an ai writing tools and document digitization platform — blog outline generator, ai paragraph rewriter, image-to-text ocr, and handwriting digitization. that product has been retired. these tools are now widely available through chatgpt, claude, gemini, and others.

shrp.app has been rebuilt from the ground up as a speech technology platform by digiwares.

rest api and mcp access is now available on starter and pro plans only. free and cloud accounts can continue using all tools through the web interface with no change to their existing limits.

signed-in users can now run smart actions directly from the live speech-to-text tool and optionally save the live transcript source to history before saving generated outputs. cloud users also get speaker diarization on file transcriptions.

users asked for a way to save live speech transcripts and refer back to them later. we built it — but in a way that stays true to how shrp works: nothing is saved unless you choose to save it.

when you run a smart action on a live transcript, shrp will ask if you want to save it to your history. you can save once, always save automatically, or continue without saving. if you don't save, the transcript stays only in your browser tab as always. saved transcripts appear in history linked to the smart action outputs they produced — so you can always see what generated what.

privacy-first means you control what gets saved — not that nothing can ever be saved.

you can now sign in to shrp using your google or github account — no password needed. one click on the auth page and you are in.

after recording a live transcript on the speech-to-text page, you can now turn it into polished audio in one click — pick a voice, preview it, and download an mp3.

• →"create audio" button added to the speech-to-text toolbar — available after any live recording.
• →ai polish step runs automatically in the background: adds punctuation, fixes casing, repairs speech-recognition run-ons.
• →optional filler word removal — removes um, uh, er from the transcript before generating audio.
• →"view original" toggle shows a before/after comparison of the polished text.
• →full voice picker inside the modal — 61 languages, multiple voice families, pre-generated mp3 previews at zero quota cost.
• →quota is drawn from your existing text-to-speech allowance — no separate limit.
• →also available on the live transcribe tab in the dashboard.

• →anonymous visitors: 500 characters per generation, 10,000 character lifetime allowance.
• →free accounts: 500 characters per generation, 10,000 + 5,000 signup bonus.
• →pro / starter: 2,000 characters per generation from your monthly plan quota.

the pay-as-you-go credit system has been retired. shrp now offers four straightforward plans: free, cloud, starter, and pro. existing plan limits and the one-time free transcription trial remain unchanged.

your dashboard now shows a full breakdown of everything you have done — every transcription, youtube transcript, text-to-speech generation, and api call, all in one place.

• →usage details page in the dashboard — tabbed view across transcription, youtube, text-to-speech, and api usage.
• →transcription history shows filename, status, duration, and file size.
• →youtube history shows video title, channel, language, and word count with a direct link to the video.
• →text-to-speech history shows a preview of the synthesised text, voice used, and character count.
• →api call log shows endpoint, status code, and cost per call.
• →each section paginates independently — load more as needed.

• →api quotas are now enforced per account, not per key. creating multiple api keys no longer multiplies your allowance.
• →free accounts now have a 25-request lifetime cap on the youtube transcript api.
• →api limits are documented on the developers page.

• →microphone mode privacy description updated to be accurate — shrp does not upload or store microphone audio. speech recognition runs through your browser's built-in speech engine.

shrp can now help you decide what to do after a transcript is created. generate transcript-specific suggestions, ask custom questions, and save useful outputs for later.

• →generate suggested actions for transcripts. shrp can recommend useful next steps based on the current transcript.
• →ask about this transcript using a custom prompt. create exam questions, explain concepts simply, extract action items, suggest what to study next, write emails, and more.
• →generated outputs can be saved and reopened later from your dashboard and history.
• →reopened youtube transcripts now support the same transcript actions workflow.

• →uploaded audio and video transcription results now support transcript actions.
• →reopened speech-to-text and file transcription history items now support suggested actions and custom prompts.
• →transcript actions adapt to the content, including meetings, lectures, interviews, and voice notes.

• →suggestions are generated from the current transcript only.
• →suggestions are button-triggered, not automatic.
• →suggestions themselves are not saved.
• →shrp does not build user profiles or long-term memory from transcript content.

• →speech-to-text now highlights common workflows including meetings, lectures, voice notes, and audio transcription.
• →homepage previews better demonstrate transcription workflows, transcript actions, and saved outputs.
• →desktop navigation was simplified with a new tools menu and clearer product structure.
• →mobile navigation was redesigned around a cleaner workflow-based layout.
• →desktop developer navigation now includes clearer api and mcp descriptions.
• →pricing page layout was improved for larger screens.

• →webhook support for transcription and transcript workflow events.
• →automation integrations built on top of webhooks.
• →transcript actions for additional workflows.
• →continued improvements to saved work, history, and generated outputs.

shrp now makes it easier to save, reopen, and continue work across transcripts, summaries, notes, and generated content.

• →youtube transcripts are now saved for signed-in users. extract a transcript once and reopen it later from your dashboard or history.
• →ai-generated outputs are now saved. summaries, key takeaways, study notes, blog drafts, linkedin posts, and x threads can be reopened anytime.
• →dashboard now has a continue working section. pick up where you left off without searching through old pages.
• →history is now more complete. speech-to-text, text-to-speech, youtube transcripts, smart tools, and saved outputs now live in one place.
• →saved outputs now support copy, txt download, markdown download, source transcript reopening, and opening the original youtube video.

• →homepage refreshed around the full workflow: capture -> transcribe -> transform -> save -> continue -> automate.
• →youtube transcript messaging is clearer and now refers to the available transcript instead of implying visible captions or cc must exist.
• →dashboard and history labels are more consistent and easier to scan.
• →api, mcp, sitemap, llms.txt, and share metadata were updated for better discovery.

• →api keys — generate shrp_live_ keys from /dashboard/api-keys. free accounts get 2 keys, paid accounts get 10. keys are shown once and hashed — never stored raw.
• →/api/v1/transcribe — speech-to-text api for audio/video files. bearer auth, plan limits, quota accounting, and direct json transcript response.
• →/api/v1/youtube-transcript — extract transcripts from any youtube video via api. bearer auth, per-plan daily limits (free 5/day · cloud 10/day · starter 30/day · pro 50/day), no quota used.
• →/api/v1/tts — convert text to speech via api. uses your existing monthly character quota. same quota as the web ui — no double billing.
• →/developers — full api docs with curl, javascript, and python examples.

• →mcp server — use shrp natively inside claude desktop, cursor, and windsurf. add one config block and your ai agent can generate speech and pull youtube transcripts.
• →tools: youtube_transcript, text_to_speech. uses the same api key and quota as the rest api.
• →endpoint: https://shrp.app/api/mcp — compatible with any mcp client.