This sequencer is being revised in revision 2 as a target for my embedded Rust audio processing framework, Rythm. Revision 2 will be a carrier board for one of the Rythm modules, and use the Rythm engine for the audio and sequencing.
 |
 |
|---|
This repository contains my project for an STM32H7-based sequencer, synthesizer, and sampler, all in a small phone-sized pocketable form factor.
- 480 MHz Arm® Cortex®-M7 32-bit Processor (
STM32H743ZIT6)- Floating-point unit
- DSP instructions
- High-speed memory controller
- Quad SPI interface for flash
- High-speed USB host and device mode
- Several SAI audio blocks
- 2.8" 240x320 TFT IPS Display (
ER-TFT028A2-4) - 24 bit Audio Codec (
NAU88C22YG)- DAC: 94dB SNR
- ADC: 90dB SNR
- 1W 8Ω BTL speaker driver
- 40mW 16Ω headphone driver
- Optional Bluetooth and Wifi (
nRF5430)- Wirelessly connect Bluethooth headphones and speakers
- Supports various high-fedality audio protocols
- BLE MIDI
- Wifi sample management and transfer
- Interal stereo 8W speakers speaker
- Internal MEMs microphone
- SDRAM for sample pool and audio processing
- Part allocated to in-memory sample pool for project
- Part allocated to audio effects processing
- Variable sizes in similar footprint available, up to 512Mbit:
- 16bit * 48000Hz = 768,000b/s = 0.768Mb/s
- 128Mb / 0.768Mb/s = 166.66 seconds of mono audio (
IS42S16800J-7TL) - 256Mb / 0.768Mb/s = 333.33 seconds of mono audio (
IS42S16160J-7TLI) - 512Mb / 0.768Mb/s = 666.66 seconds of mono audio (
IS42S16320F-7TLI)
- Micro SD Card
- Saving and loading projects
- Transfer and back up projects to a computer
- Saving and loading samples
- Load samples from a computer
- Save samples recorded on the device
- Saving and loading projects
- 3.3v NOR Flash (
MT25QUxxxxxx1xW7256M/128M/64M)- Stores device settings
- Small storage space for projects and samples without SD card
- Provides storage for cross-SD card samples
- USB-C
- USB-PD fast charging
- Possible future audio interface mode
- USB MIDI in and out
- MIDI in/out via TRS minijack type A
- 16 Multi-Purpose primary keys
- Hot-swappable Kalih V1 Choc keyswitches
- Addressable RGB LEDs for state indicators
- Accelerometer and Gyroscope (
LSM6DS3)- Can be used as sequencer parameters
- Can be recorded for automation inputs
The step sequencer on the device is modeled after popular semantics used by hardware sequencers, such as the Elektron Digitone and Digitakt, OP-1, OP-Z, Pocket Operators, Woovebox, etc.
The hard requirements of the step sequencer and audio engine are not defined yet because they are under active development.
- As many projects as the SD card or internal memory can store
- 128 Patterns (TBD)
- 8-16 Audio or MIDI tracks per pattern (TBD)
- 8-24 Songs (arrangements of patterns) (TBD)
- 128 Patterns (TBD)
- 8-16 voice polyphony (TBD)
- 2-3 LFO per voice (TBD)
The primary audio engine runs in 32-bit and supports stereo audio.
The sampling engine runs 16bit 44.1Khz and saves and loads samples in single-channel mono.
Samples recoded from the internal or external mic are sampled at 24 bits into the device's SDRAM to provide a decent noise floor, and then filtered and downsampled to 16 bits for saving after they've been edited by the user.
The 16bit @ 44.1Khz sampling parameters provdes a good balance between audio quality, and reasonable utilization of the SDRAM without needing extremely large and expensive SDRAM chips:
16bit * 48000Hz = 768,000b/s = 0.768Mb/s
128Mb / 0.768Mb/s = 166.66 seconds of mono audio
256Mb / 0.768Mb/s = 333.33 seconds of mono audio
512Mb / 0.768Mb/s = 666.66 seconds of mono audio
Note that some of the space calculated above (actual amount will be configurable) is allocated to the audio effects processing chain, i.e. for reverb and delay.
- HardFault: Under XPERIPHERALS in Cortex-debug check the
CFSR_UFSR_BFSR_MMFSRregister to see the fault code and reference: https://community.st.com/t5/stm32-mcus/how-to-debug-a-hardfault-on-an-arm-cortex-m-stm32/ta-p/672235#toc-hId--1471762501