Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed the week of November 23.

TLV320AIC3254: audio codec processing capability

Part Number: TLV320AIC3254

Hi Sir,

My customer is currently evaluating whether the TLV320AIC3254 meets their requirement for an audio codec with DSP functionality in their DSC products. Could you please help confirm if the following features are supported?

Application: DSC
Input: 2x PDM DMICs (Stereo) | SNR: 64 dBA | Sensitivity: -26 dBFS
Output: 1x Speaker (0.4W, 10phi)
* Note: The audio output can also be processed by the SoC to support USB UAC (USB Audio Class) output.

I. miniDSP Customization:
Can the TLV320AIC3254's built-in miniDSP be programmed to customize the processing block routing sequence for input audio, as shown in the diagram below?

II. DMIC Input Audio Processing:
1. Input Level Adjustment (ADJ)
After the DMIC input, can the TLV320AIC3254 adjust the input signal level prior to audio processing to optimize the signal?

2. Sound Quality Correction (EQ)
Regarding the Low-Cut Filter (LCF) / High-Pass Filter applied to the input audio signal: Can the filter be configured as a 1st-order filter? Can the cutoff frequency be set to 100Hz?

3. Noise Reduction (NR)
Can Wind Noise Reduction (WNR) be implemented using the built-in DSP?

4. Level Adjustment (LEVEL)
Can the output level be adjusted via Gain Control with configurable ranges and step sizes?
Can the left and right channels be adjusted synchronously (stereo-linked) to prevent stereo image shift?
When the input volume is too high, can AGC or a Limiter be applied to prevent output clipping?

5. Automatic Gain Control (AGC)
Can the AGC compress the dynamic range of the microphone input signal without causing unnatural audio artifacts, while raising the average volume to improve audibility?
Do the static characteristics include Limiter, Compressor, and Expander (Noise Gate) functions, and do the dynamic characteristics include Attack and Release settings?
If the built-in AGC is insufficient, can additional DSP blocks be implemented to supplement it?

III. Speaker Output Audio Processing:
1. LR MIX
Can the stereo audio be mixed down to mono, and can this routing be restricted specifically to the media playback path?

2. EQ
Can a Parametric EQ or IIR Filter be used to adjust the frequency response of the output audio?
Is it possible to configure separate EQ parameters for media playback, camera operation sounds, and UI/operation sounds?

3. Volume Adjustment (LEVEL)
Can independent volume settings for different audio sources be achieved via software control or parameter switching?

4. AGC
Can AGC processing be applied to the mono audio after the LR MIX?
Are the AGC parameters adjustable (e.g., Thresholds for the Limiter/Compressor/Expander, Ratio/Slope for the Compressor/Expander, and Attack/Release times)?

IV. DMIC SNR Performance:
Is it possible to improve the DMIC SNR performance to >45 dBA?

V. Stereo DMIC Placement:
Regarding the stereo DMIC L/R inputs, is there a recommended placement distance (spacing) between the left and right DMICs to achieve an optimal stereo image when using the integrated audio codec?

VI. USB Audio Routing:
Can the audio output also be routed through the SoC to support USB UAC (USB Audio Class) output?

 

 

  • Hello,

    Our expert on this device is out of office, please allow our team until next week to respond.

    Best,

    Garret

  • Hi,

    Most of these steps can be achieved with the miniDSP on the device, which can be programmed with PurePath Studio software. The main step that needs more care is the wind noise reduction - we do not have a specific wind noise reduction algorithm built, but you can use the SS_NR object, or the AGC with the DRC, or whatever combination you choose.

    All features of the miniDSP are register writes, but they are to hidden registers and mappings are all handled by the PurePath Studio software. To change the volume or EQ for a specific signal sent out by your MCU, you will have to handle the register writes to change the controls in that case. 

    The minimum best-case SNR of the ADC in the AIC3254 is 80dB, according to the datasheet. It will typically be between 80 and 90dB, A-weighted. In a product with the digital mic on, the system's SNR will be limited to the performance of the microphone. Placement of microphones is up to you, this is not related to the codec performance. 

    Most of our audio EVMs include a USB connection which can provide I2C communication and audio data transfer over USB. The AIC3254 EVM should include the AC-MODEVM, which has an XMOS MCU with our driver that handles an ASIO USB audio protocol. You are welcome to use your own MCU to handle audio over USB in your final product.

    Here is a mockup of the steps you outlined here for the signal flow for both ADC and DAC in the miniDSP. Let me know what you think or if you have more questions.

    Best,
    Mir

  • Hi Sir,

    The customer has brought up the following inquiries. Could you please assist with a reply? Thank you.

    Input Adjustment (ADJ)
    Is this parameter adjustable? If so, what is the approximate adjustable range?

    Sound Quality Correction (EQ)
    Is this configured via frequency bands? If so, how many bands are available for configuration?

    External Amplifier Connection
    For output: 1x Speaker (0.4W, 10phi), can the codec connect to an external audio amplifier, such as the LM4675SDX/NOPB or TPA2011D1YFFR ?

    Wind Noise Reduction (WNR)
    Regarding the previously mentioned SS_NR or the AGC + DRC combination, are these configured directly within the PurePath Studio software?

    Noise Reduction (NR) Performance
    Do you have empirical data or example test results showing the typical noise reduction performance (e.g., the noise floor is reduced from X dB to Y dB)?

    Customer NR Specifications
    Can the DSP meet the following noise reduction specifications provided by the customer?

    Wind Noise: When a 2.0 m/s airflow is blown against the camera body, and the angle is adjusted to maximize the noise level, the calculated input sound pressure equivalent must not exceed:
    Wind Noise: Less than or equal to 60 dBA (with LCF enabled)
    Zoom Noise: Less than 2.5 dB with no audio artifacts or abnormal noise.
    Autofocus (AF) Noise: Less than 2.5 dB with no audio artifacts or abnormal noise.
    Camera Shake (OIS) Noise: Less than 1.5 dB with no audio artifacts or abnormal noise.


    Level Adjustment (LEVEL)
    Can the signal level be increased or decreased within a step range of 0 to 31, with the reference level set to 26?

  • Hi,

    Input Adjustment (ADJ)
    Is this parameter adjustable? If so, what is the approximate adjustable range? -> the ADC volume can be adjusted with I2C between -12 and 20dB with .1dB steps, and in the PPS GUI, the digital volume can be adjusted -110 to 6dB in increments of 1dB

    Sound Quality Correction (EQ)
    Is this configured via frequency bands? If so, how many bands are available for configuration? -> yes, we offer biquad filters or FIR filters in PPS. I tried an example with 11 biquad filters and was only using 90% of available ADC cycles. You can use as many bands as there are available DSP cycles left in the processor, it depends on the rest of the PPS flow.

    External Amplifier Connection
    For output: 1x Speaker (0.4W, 10phi), can the codec connect to an external audio amplifier, such as the LM4675SDX/NOPB or TPA2011D1YFFR ? -> yes

    Wind Noise Reduction (WNR)
    Regarding the previously mentioned SS_NR or the AGC + DRC combination, are these configured directly within the PurePath Studio software? -> yes

    Noise Reduction (NR) Performance
    Do you have empirical data or example test results showing the typical noise reduction performance (e.g., the noise floor is reduced from X dB to Y dB)? -> No, we cannot reduce the noise with noise cancellation algorithms since our device does not have the processing power for noise cancellation, just can reduce noise with filtering and compression and noise gate. So, the noise reduction depends on what you count as noise and what you count as reduction... this is dependent on the end system and subjective.

    Customer NR Specifications
    Can the DSP meet the following noise reduction specifications provided by the customer? -> They would need to do their own testing, it depends on the input signal. We do not have specific metrics we say we can hit with these noise reductions, it would have to be tuned by the customer with their mic input signal

    Wind Noise: When a 2.0 m/s airflow is blown against the camera body, and the angle is adjusted to maximize the noise level, the calculated input sound pressure equivalent must not exceed:
    Wind Noise: Less than or equal to 60 dBA (with LCF enabled)
    Zoom Noise: Less than 2.5 dB with no audio artifacts or abnormal noise.
    Autofocus (AF) Noise: Less than 2.5 dB with no audio artifacts or abnormal noise.
    Camera Shake (OIS) Noise: Less than 1.5 dB with no audio artifacts or abnormal noise.


    Level Adjustment (LEVEL)
    Can the signal level be increased or decreased within a step range of 0 to 31, with the reference level set to 26? -> I don't know what this means but the levels are adjustable by .1dB step size, as specified in before answer.

    Let me know of any more questions.

    Best,
    Mir