Type-to-speak app in Indian languages for adults who lose their voice
TL;DR Typespeak is a phone and tablet app for literate adults who can no longer speak, mainly after oral or throat cancer surgery or with ALS. They type in Hindi, Tamil or another Indian language and the app predicts phrases, then speaks in a copy of their own voice recorded before surgery. Cancer hospitals refer patients and families pay.
Oral cancer is the most common cancer among Indian men, and it's brutal on speech. The IARC's India fact sheet estimates 205,878 new lip and oral cavity cancers a year, plus 35,169 of the larynx[1]. Surgery for some of these removes much of the tongue or the voice box, and the patient wakes up unable to talk to their family.
So what can they use? Not much that was built for them. Indian communication apps are mostly picture boards designed for children. The leading text-based app for adults offers European and a few other languages, with no Indian one listed[2]. Indian speech therapists say high-tech aids are held back by cost, training and a lack of local guidelines[3][4]. So a lot of patients end up writing on paper and pointing.
What it does, in regulatory terms: a communication aid only. Typespeak does not diagnose, monitor or treat anything. CDSCO's definition of a medical device includes software for "assistance for any injury or disability"[5], so get a written regulatory opinion before launch. The app may need registration as a low-risk device.
Here's the flow. Before surgery a speech therapist records the patient reading a short passage. Typespeak builds a synthetic copy of that voice. After surgery the patient types in Roman script or their own script and picks from phrases predicted from their own messages, and the phone speaks. A quick-tap screen covers ward needs such as pain, suction and calling a nurse. Speech runs on the device, so audio of the patient's voice never leaves it.
The seed pitched predictive AAC for everyone with complex communication needs, which I think is too wide. One Indian AAC app already sells picture-and-text AAC in six Indian languages[6], and a free academic app covers Hindi and English[7]. Literate adults who lose speech late in life are the ones left out.
The patient uses it but the family usually pays, often right after a large surgical bill. Today the choices are the leading Indian AAC app at ₹11,999[6] or an English-first type-to-speak app at a US$149.99 list price, about ₹14,500[2]. I'd test a lower annual subscription, though that price is an assumption for now. Cancer hospitals are the second payer, buying ward licences for patients in recovery.
The voice tech became free and Indian, which is a big deal here. An open speech model from an Indian university lab speaks 11 Indian languages and can copy a voice from a short reference clip[8]. In October 2026, a Bengaluru cancer hospital reported cloning 26 patients' voices before tongue surgery from under a minute of audio each, with patients and families rating identity at about 4.5 out of 5[9]. CDSCO's 2026 software guidance makes the device question answerable[5].
There's no public figure for adult AAC in India, and I couldn't find a count of glossectomies or laryngectomies either. The anchors are 205,878 new oral cavity and 35,169 larynx cancers a year[1], of which only a share lose speech. ALS adds a smaller group: in one Indian caregiver survey, 43.8% of 130 patients had onset in the speech and swallowing muscles[10].
A Chennai AAC maker leads the Indian market with picture and text modes and predictive text, mostly for children[6]. A free academic app offers icon boards[7]. A European type-to-speak app is the closest model for adults: type-to-speak with prediction and own-voice support, but no Indian languages[2]. The Bengaluru voice study is research, not a product[9]. Nobody puts Indian-language typing, phrase prediction and a saved own voice together. That's the gap.
| Difficulty | Moderate (41/100) |
| Starting capital | ₹85 L–1.55 Cr |
| Time to launch | 6–9 months |
Prediction and speech in Indian languages work now, but tuning them for typed Hinglish and slow typists takes time. The harder bits are getting into the pre-surgery workflow, which means persuading surgeons and speech therapists at busy hospitals, and settling whether the app is a medical device.
This one may need a licence. Whether it does depends on how CDSCO reads it.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
AI is most of the product here, in three places.
Don't let AI speak for the patient. Predictions are suggestions the person confirms, and the app does not diagnose, monitor or treat anything.
Want to work out the AI layer for Typespeak? Talk to Yash →
I like this for a founder with speech-tech or Indian-language NLP experience who has a partner inside a cancer hospital. If you're hoping for a mass consumer app then skip it. Prove two things first - one hospital records voices before surgery for 30 patients, and 20 of them still use the app a month later.
Plan for roughly ₹85 L–1.55 Cr, covering speech and prediction work, a regulatory opinion, possible device registration and hospital partnerships.
Possibly. CDSCO's device definition covers software that assists with a disability, so get a written opinion. It may need low-risk device registration.
Yes, wherever adults lose speech and their language is poorly served by existing apps. Check your own medical-device and voice-data rules first.
The free Typespeak validation kit has the line-by-line capital breakdown, the effort scoring, the device and voice-data checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for patients, families and hospitals, and 5 assumptions to test. Download the kit →
Thinking about building Typespeak or something close to it? The make-or-break bit is getting voice recording into the hospital's routine before surgery, and that's exactly the kind of thing I help founders with. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.