Low Power, Wideband Voice Processor

Introduction
Fortemedia’s FM34, an innovative low power wideband voice processor, is an excellent choice for portable applications such as mobile phones and handhelds. Fortemedia’s SAM VE technology embedded in the processor and denoted by the suffix, as in FM34 supersedes conventional array microphone design limitation with more flexibility in microphone placement.

FM34 provides exceptional noise suppression, echo cancellation and side tone generation delivered in a small, self contained SOC. It combines digital microphone interface, 8/16 KHz sampling rate and digital signal processing to achieve high fidelity wideband performance.

Overview
The FM34 includes a high performance and low power DSP that comes with hardware accelerators, internal ROM and RAM and a DAC that can be configured to be used on the transmit or receive path. When put on the receive path, it is capable of driving directly standard 32ohm receiver of the handset. On the transmit path, the SOC can be connected to the host system baseband either through analog microphone input or digital audio interface- I²S or PCM. The microphone array input is captured through digital microphone interface, passed through FM34 digital voice processor and the processed data output to the host system.

Alternatively, voice samples digitized by a codec can be fed through the I²S interface instead of the PDM input of the processor. The new Bright Voice processing technology enhances the playback intelligibility of the far side voice by monitoring the local ambient noise and enhancing the speaker output. The FM34 allows flexibility for microphone placement inside any mobile devices. Its small 25-pin WLCSP package/32-pin QFN package and the extremely low power consumption broaden its embedded application possibility. It is pin compatible with FM33 series of WLCSP processors. FM34 accepts audio input from Fortemedia Digital Array Microphone or other manufacturers’ digital microphones.

Key Features
Hardware Advantages:
● Highly-integrated mobile voice processing subsystem
● Customized high performance 16-bit voice processor with built-in hardware accelerations
　for 8/16 kHz sampling rate signal processing. Supports various microphone configurations
　and feature set combinations.
● 1 DAC (Digital-to-Analog Converter) for line-out output, 90dB SNR and differential output
　to eliminate RF interference.
● DAC sampling rates: 8k, 16k, 32k, 44.1k, 48kHz
● DAC/Headphone output start up time < 200ms
● DAC/Headphone driver PSRR >70dB
● One headphone PA to drive the 32ohm receiver for handset application
● Pop suppression circuits for both DAC and headphone outputs
● Software controllable separate line level and headphone DAC output pins
● Programmable DAC and headphone output impedance
● I²S or PCM digital interface for audio data transmission between host system and
　audio codec
● Two synchronous PCM/I²S interfaces
● Mono PCM port mode supports 8-bit A-Law, μ-Law, 13-bit zero padding PCM and
　16-bit linear PCM formats.
● Programmable digital gain control from +24dB～-143.25dB with 0.75dB increments for
　input and output voice port
● Supports 8/16kHz voice processor bypass mode including
　8/16/24/32/44.1/48kHz PCM/ I²S interfaces
● PCM output high impedance when PCM disabled.
● Sampling microphone inputs at 16-bit 16kHz
● Covers wideband voice to 7250Hz
● 2 decimators for converting digital microphone PDM (Pulse Density Modulation) to PCM
● Built-in PLL (Phase-Locked Loop) supports master clock source of 3MHz～32MHz in
　1MHz incremental steps or 4.096MHz～32.768MHz in 2.048MHz incremental steps.
● Popular system clocks - 3.6864, 7.68, 14.4, 15.36, 16.8, 19.2, 19.68, 38.4 and 48MHz
　are also supported
● Host controls FM34 using SHI (Serial Host Interface) bus, speed up to 400kbps.
● DMIC interface clocks 1.024.
● Low-power consumption with power down mode support
● 110nm advanced process and low active power consumption.
　• Typical power consumption is around
　　- 18mW in 2 microphones mode including DAC, 1.2V core operation
　　　and headphone driver off
　　- 12μW in power-down mode.
　• 3.3V for analog power supply
　• Native 1.2V for digital and core blocks or use 1.8V or higher VDDIO through
　　on-chip regulator
　• 1.8V to 3.3V tolerant I/O pins
　• Parameter retention after resuming from power-down mode: No reprogramming needed
　• Wakeup time is less than 10ms
● Small footprint package - 25-pin WLCSP package with 0.45mm pitch
● Low power HW bypass modes: PCM/ I²S interfaces act as a pass through switch in
　bypass mode
● SW bypass modes:
　• Acts as a plain codec or ADCs while voice processing is bypassed
　• 16-bit resolution recording with rates up to 48kHz, including popular 8, 16, 32, 44.1
　　and 48kHz
　• DMIC → I²S stereo
　• One of the I²S channels → DAC
● Low-power consumption with power down mode support

Software Advantages:
● 8/16kHz sampling rate and processing meet wideband bandwidth requirement
　• Narrow band voice 180Hz～3400Hz bandwidth based on 8kHz sampling rate
　• Wideband voice 120Hz～7400Hz bandwidth based on 16kHz sampling rate
● Supports omni-omni microphone array with front/back placement, 6～14cm distance apart
　and at least 6dB receiving signal difference for SAM VE to create close-talk voice
　extraction zone
● Supports single microphone operation
● Software programmable VE cone angles through parameters settings
● Superior full duplex double-talk operation
● Adaptive step-sized sub-band process with dedicated non-linear filter delivers
　AEC (acoustic echo cancellation) performance up to 70dB
● Suppresses non-stationary noise up to 37dB and stationary noise up to 35dB
● Acoustic echo tail length: up to 100ms
● DRC enhances voice intelligibility and full duplex
● New Bright Voice speech enhancement technology improves local speaker output
　intelligibility
● Receiving path noise suppression (Rx NS) reduces far-end user’s background noise
● Side tone generation
● SHI supports on-the-fly commands and registers programming
● No reprogramming needed after resuming from power-down mode

▲TOP