Skip to the content
2 layers100.8 × 91 mmOutHouse, the printed circuit board in 3D

HHD3.0

Size
100.8 × 91 mm
Copper layers
2
Thickness
1.6 mm
Parts
10
Nets
9
Pads
84
Tracks, vias
78, 1
Routed
Yes
KiCad
KiCad 9.0 (file 20241229)
Created
12 Apr 2026
Updated
12 Apr 2026

Overview

OutHouse is an open-source printed circuit board by BharathKrishna0078, from the BharathKrishna0078/Vaacha repository on GitHub. It is a 2-layer board of 100.8 × 91 mm with 10 parts on 9 nets, routed with 78 tracks and 1 via.

README

README.md

VAACHA — Indian Sign Language to Multilingual Speech Translator

Vaacha (वाचा) = "speech/voice" in Sanskrit — giving voice to the voiceless


What Is VAACHA?

VAACHA is a handheld hardware device that enables two-way communication between deaf/hard-of-hearing individuals and hearing people — across any Indian language — without an interpreter.

DirectionFlow
Deaf → HearingISL gesture → TinyML on ESP32-S3 → Sarvam TTS → spoken audio
Hearing → Deafspoken audio → Sarvam STT → Sarvam Translate → OLED text display

One device. Two-way communication. 11 languages. No cloud required for gesture recognition.

See VAACHA-PRD.md for the full Product Requirements Document.


The Problem

18 million deaf and hard-of-hearing Indians cannot communicate independently at doctor visits, bank counters, railway ticket windows, or any daily transaction — because there are only ~300 certified ISL interpreters in all of India.

Existing tools fail because India has 22 official languages. Even if a device "understands" ISL and outputs text, it can't bridge a deaf Tamil woman speaking to a Hindi-speaking doctor.


Hardware Architecture

The device runs on two ESP32 modules mounted on a custom-designed carrier PCB:

BoardRole
ESP32-CAM (OV2640)Captures gesture frames → sends via UART to Glyph S3
Glyph S3 (ESP32-S3)Main brain — TinyML inference, audio I/O, OLED, buttons, WiFi to laptop

The ESP32-S3 connects to a laptop over local WiFi (no internet) via WebSocket. The laptop runs the Sarvam AI pipeline.

text
[VAACHA Device]                         [Laptop Hub]
  ESP32-CAM                              Python App
    └─UART→ Glyph S3 (TinyML)              ├── Sarvam STT
               └──── WiFi/WebSocket ──────► ├── Sarvam TTS
                                            ├── Sarvam Translate
                                            └── Flask Dashboard

Custom Carrier PCB

The PCB (70mm × 50mm, 2-layer, KiCad 8) integrates:

  • Mic bias circuit (2.2K + 100nF) for clean analog audio
  • Speaker amplifier stage (RC filter + BC547 transistor)
  • LiPo → 3.3V LDO power management
  • 4× M2.5 mounting holes for the 3D-printed enclosure
  • JST connectors for battery and speaker

Without the PCB: two dev boards on a breadboard.
With the PCB: a single integrated product you can hold in your hand.


Software Architecture

TinyML Gesture Classifier (On-Device)

ParameterValue
FrameworkTensorFlow Lite Micro (via Edge Impulse)
ModelMobileNetV2 INT8 quantized
Input96×96 grayscale image
Output15 ISL gesture classes + confidence
Model size~200–300 KB
Inference time200–400 ms
Confidence threshold0.75

Hackathon Gesture Vocabulary (15 signs)

#GestureMeaning
1Open palm waveHello
2Thumbs upYes
3Thumbs downNo
4Both palms upHelp
5Point to selfMe/I
6Flat hand to mouthFood/Eat
7Cupped hand to mouthWater/Drink
8Fist to chestPain/Hurt
9Cross armsStop
10Palm forward pushGo/Leave
11Point forwardYou/There
12Finger circleMoney/Pay
13Roof shapeHome/House
14Hand to earListen/Hear
15Open palm forwardThank you

Laptop Hub (vaacha-hub/)

text
vaacha-hub/
├── main.py              # Entry point — WebSocket server + dashboard
├── sarvam_pipeline.py   # Sarvam STT, TTS, Translate, Language Detection
├── ws_server.py         # WebSocket server (receives from ESP32 via WiFi)
├── audio_utils.py       # ADPCM decode/encode, WAV conversion
├── dashboard/           # Flask live demo visualization
├── config.py            # Language codes, API keys, WiFi settings
└── requirements.txt     # sarvamai, websockets, flask

Sarvam AI Integration

Sarvam APIUsed For
Text-to-Speech (/text-to-speech)Gesture label → spoken audio in target language
Speech-to-Text (/speech-to-text)Transcribe hearing person's speech
Translation (/translate)Cross-language text conversion
Language Detection (/detect-language)Auto-detect what language was spoken
Transliteration (/transliterate)Display text in user-preferred script

Supported output languages: Hindi, Tamil, Telugu, Kannada, Malayalam, Bengali, Marathi, Gujarati, Punjabi, Odia, English


Operating Modes

ModeLED ColorWhat Happens
SIGN🟢 GreenCamera active → gesture classified → Sarvam TTS → speaker
LISTEN🔵 BlueMic active → Sarvam STT → translated text on OLED
IDLE⚪ White pulseLow power, waiting for button press

Toggle modes with a single button press. Language cycles with the other two buttons. Preference is saved to flash.


Bill of Materials

ComponentPurpose₹
Glyph S3 (ESP32-S3)Main MCU + TinyML + WiFi2,399
OV2640 Camera (ESP32-CAM)Gesture capture(from organizers)
1.3" I2C OLED (SH1106)Text display499
Condenser MicAudio input for STT20
Speaker 0.5W 8ΩAudio output for TTS150
LiPo BatteryPortable power300
Tactile buttons × 3Mode + language select30
RGB LEDs × 2Status indicator20
BC547, resistors, capsAmp stage + filters + bias105
Headers, wires, perfboardAssembly110
TOTAL₹3,633

₹367 buffer remaining from ₹4,000 hackathon budget.


Repo Structure

text
Hardware Hackathon/
├── README.md           ← You are here
├── VAACHA-PRD.md       ← Full Product Requirements Document
└── PCB/                ← KiCad schematic + layout files

Hackathon Track

LOCALAI + HARDWARE — Edge AI running on the MCU (TinyML gesture classification on ESP32-S3), local AI on the laptop (Sarvam Edge models, offline), no cloud required for core functionality.


Future Roadmap

PhaseTimelineMilestone
v1.0Hackathon (3 days)15 gestures, laptop hub via WiFi, Sarvam Cloud API
v1.5+2 months50 gestures, fully offline with Sarvam Edge on laptop
v2.0+6 monthsAndroid companion app with Sarvam Edge + BLE — no laptop needed
v2.5+9 months200 gestures, iOS app, dynamic gesture sequences
v3.0+12 monthsAICTE/NCERT partnership for deaf schools, multilingual ISL dictionary

Team Roles

RoleResponsibility
Hardware LeadKiCad PCB design, soldering, analog circuits
Firmware DevESP-IDF / Arduino, FreeRTOS tasks, camera + audio I/O
ML EngineerEdge Impulse training, TFLite model deployment
Software DevSarvam pipeline, WebSocket server, Flask dashboard

Parts

10
QtyValueFootprintReferences
1U1glyph-s3U1
1U10R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_HorizontalU10
1U2OLED_128X64_1.3U2
1U3ESP 32 CAMU3
1U4SW_PUSH_6mmU4
1U5SW_PUSH_6mmU5
1U6gmod_lithium_charging_moduleU6
1U7gmod-dc-dcU7
1U8JST_EH_S2B-EH_1x02_P2.50mm_HorizontalU8
1U9R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_HorizontalU9

Similar boards

  • 2 layers57 × 55.2 mmLogicAnalyzer, the printed circuit board in 3D

    LogicAnalyzer

    gusmanb/logicanalyzer

    24 channel, 100Msps logic analyzer hardware and software

    10 parts5.1kGPL-3.0
  • 2 layers32.2 × 15.4 mmProtector, the printed circuit board in 3D

    Protector

    byrantech/laptop

    anyon_e, a highly integrated, high end, open source laptop. Attempt the impossible.

    10 parts4.3kMIT
  • 2 layers26 × 18 mmConn, the printed circuit board in 3D

    Conn

    yaqwsx/KiKit

    Automation tools for KiCAD

    10 parts2.1kMIT
  • 2 layers44.3 × 30.5 mmLime2-SATA RevA, the printed circuit board in 3D

    Lime2-SATA RevA

    OLIMEX/OLINUXINO

    OLINUXINO is Open Source / Open Hardware, low cost from EUR 24 Linux Industrial grade Single Board Computer capable to operate -25+85C

    10 parts1.5k
  • 2 layers17.8 × 43.2 mmNano Adapter, the printed circuit board in 3D

    Nano Adapter

    VoronDesign/Voron-Hardware

    Cool Hardware designed by the Voron Design group

    10 parts1.1kGPL-3.0
  • 2 layers53.9 × 23.2 mmBuffer, the printed circuit board in 3D

    Buffer

    hoglet67/RGBtoHDMI

    Bare-metal Raspberry Pi project that provides pixel-perfect sampling of Retro Computer RGB/YUV video and conversion to HDMI

    10 parts1kGPL-3.0

Questions

How big is the OutHouse PCB?

It measures 100.8 × 91 mm and has 2 copper layers, 1.6 mm thick.

What parts does the OutHouse use?

10 parts. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the OutHouse KiCad files?

From the BharathKrishna0078/Vaacha repository on GitHub, at the commit this page was made from: the board is hardware/pcb/OutHouse.kicad_pcb. PCBKit opens the board in the browser, with nothing to install.

Can I use the OutHouse design in my own project?

Its repository states no licence, so its author keeps all rights: ask BharathKrishna0078 before reusing the design.