Introduction: Play Video With ESP32

This Instructables show something about playing video and audio with ESP32.

Step 1: ESP32 Features & Limitations


  • 4 SPI bus, 2 SPI bus available for user space, they are SPI2 and SPI3 or called HSPI and VSPI. Both SPI buses can run at most 80 MHz. Theoretically it can push 320x240 16-bit color pixels to SPI LCD at 60 fps, but it has not yet counted the time overhead required for read and decode the video data.
  • 1-bit / 4-bit SD bus can connect SD card in native protocol
  • I2S internal DAC audio output
  • over 100 KB RAM available for video and audio buffer
  • Fair enough processing power to decode JPEG (play Motion JPEG) and LZW data compression (play Animated GIF)
  • Dual-core version can split read data from SD card, decode and push to SPI LCD into parallel multi-tasks and boost the playback performance


  • not enough internal RAM to have double frame buffer for 320x240 in 16-bit color, it limited the multitask design. It can overcome a bit with external PSRAM though it is slower than internal RAM
  • not enough processing power to decode mp4 video
  • not all ESP32 version have 2 core, the multi-task sample only benefit on dual-core version


Step 2: Video Format


Or called 16-bit color is a raw data format commonly used on the communication between MCU and color display. Each color pixel represented by a 16-bit value, the first 5-bit is red value, following 6-bit is green value and then 5-bit blue value. 16-bit value can make 65536 color variation so it also called 64K colors. So 1 minute 320x240@30 fps video will be sized: 16 * 320 * 240 * 30 * 60 = 2211840000 bits = 276480000 bytes or over 260 MB

Animated GIF

This is a common file format on the web since 1990s. It limit the color variation for each screen up to 256 colors and do not repeat store the pixel that as same color as previous frame. So it can much reduce the file size, especially when each animation frame not change too much details. The LZW compression is designed capable decoded by 1990s computer, so ESP32 also have fair enough processing power to decode it in real time.

Motion JPEG

Or called M-JPEG / MJPEG is a common video compression format for the video capture hardware with limited processing power. It actually simply a concatenation of still JPEG frames. Compare with MPEG or MP4, Motion JPEG no need computationally intensive technique of interframe prediction, every frame is independent. So it requirement less resource to encode and decode.


Step 3: Audio Format


A raw data format for digital audio. ESP32 DAC use 16-bit bit depth, that means each 16-bit data represent a digital sampled analog signal. Most video and song audio commonly use sample rate at 44100 MHz, that means 44100 sampled analog signal for each second. So, 1 minute mono audio PCM raw data will be sized: 16 * 44100 * 60 = 42336000 bits = 5292000 bytes or over 5 MB. The size of stereo audio will be double, i.e. over 10 MB


MPEG Layer 3 is a compressed audio format widely used for song compression since 1990s. It can dramatically reduce file size to under one-tenth of raw PCM format


Step 4: Format Conversion

This project use FFmpeg convert the video into ESP32 readable format.

Please download and install FFmpeg at their official site if not yet:

Convert to PCM audio

ffmpeg -i input.mp4 -f u16be -acodec pcm_u16le -ar 44100 -ac 1 44100_u16le.pcm

Convert to MP3 audio

ffmpeg -i input.mp4 -ar 44100 -ac 1 -q:a 9 44100.mp3

Convert to RGB565

ffmpeg -i input.mp4 -vf "fps=9,scale=-1:176:flags=lanczos,crop=220:in_h:(in_w-220)/2:0" -c:v rawvideo -pix_fmt rgb565be 220_9fps.rgb

Convert to Animated GIF

ffmpeg -i input.mp4 -vf "fps=15,scale=-1:176:flags=lanczos,crop=220:in_h:(in_w-220)/2:0,split[s0][s1];[s0]palettegen[p];[s1][p]paletteuse" -loop -1 220_15fps.gif

Convert to Motion JPEG

ffmpeg -i input.mp4 -vf "fps=30,scale=-1:176:flags=lanczos,crop=220:in_h:(in_w-220)/2:0" -q:v 9 220_30fps.mjpeg


  • FFmpeg converted Animated GIF can be further optimizered by some web tools, you may search GIF optimizer to find one.

Step 5: Hardware Preparation

ESP32 Dev Board

Any dual-core ESP32 dev board should be ok, this time I am using a TTGO ESP32-Micro.

Color Display

Any color display that Arduino_GFX support should be ok, this time I am using a ILI9225 breakout board with SD card slot.

You can find Arduino_GFX supported color display list at Github:

SD Card

Any SD card should be ok, this time I am using a SanDisk "normal speed" 8 GB micro SD with SD adaptor.


If you want to use headphone only, simply connect headphone pins to pin 26 and GND can listen the audio. Or you can use a tiny amplifier to play audio with speaker.


Some breadboards and breadboard wires

Step 6: SD Interface

ILI9225 LCD breakout board also included a SD crd slot breakout pins. It can be used as SPI bus or 1-bit SD bus. As mentioned in my previous instructables, I prefer using 1-bit SD bus, so this project will base on 1-bit SD bus.

Step 7: Put It Together

The above pictures show the testing platform I am using in this project. The white breadboard is 3D printed, you can download and print it at thingiverse:

The actual connection depends on which hardware you have in hand.

Here are the connection summary:

Vcc     -> LCD Vcc
GND     -> LCD GND
GPIO 2  -> SD D0/MISO -> 1k resistor -> Vcc


Step 8: Program

Arduino IDE

Download and install Arduino IDE if you are not yet do it:

ESP32 Support

Follow the Installation Instructions to add ESP32 support if you re not yet do it:

Arduino_GFX Library

Download latest Arduino_GFX libraries: (press "Clone or Download" -> "Download ZIP")

Import libraries in Arduino IDE. (Arduino IDE "Sketch" Menu -> "Include Library" -> "Add .ZIP Library" -> select downloaded ZIP file)


Download latest ESP8266Audio libraries: (press "Clone or Download" -> "Download ZIP")

Import libraries in Arduino IDE. (Arduino IDE "Sketch" Menu -> "Include Library" -> "Add .ZIP Library" -> select downloaded ZIP file)

RGB565_video Sample Code

Download latest RGB565_video sample code: (press "Clone or Download" -> "Download ZIP")

SD Card Data

Copy the converted files to SD card and insert into LCD card slot

Compile & Upload

  1. Open SDMMC_MJPEG_video_PCM_audio_dualSPI_multitask.ino in Arduino IDE
  2. If you are not using ILI9225, change the new class code (around line 35) to correct class name
  3. Press Arduino IDE "Upload" button
  4. If you failed to upload the program, try detach the connection between ESP32 GPIO 2 and SD D0/MISO
  5. If you find the orientation not correct, change the "rotation" value (0-3) in new class code
  6. If program run well you can try other sample start with SDMMC_*
  7. If you do not have SD card slot or you don't have FFmpeg installed, you can still try SPIFFS_* example

Step 9: Benchmark

Here are the performance summary for different video (220x176) and audio (44100 MHz) format:

FormatFrame per second(fps)
RGB565 + PCM9


  • MJPEG + PCM can reach higher fps but it is unnecessary play in a tiny screen greater than 30 fps
  • RGB565 does not require decode process but the data size is too large and much time consumed at loading data from SD, 4-bit SD bus and faster SD card can improve it a little bit (wild guess can reach around 12 fps)
  • MP3 decode process not yet optimized, it is now dedicate core 0 for MP3 decode and core 1 for playing video

Step 10: Happy Playing!

Now you can play video and audio with your ESP32, it unlocked many possibilities!

I think I will make a tiny vintage TV later...