TLV320AIC3262: Background noise observed in handset mode

Hi,

For my reference, here is the previous forum post. 

I'm having trouble playing the audio files, but I see the I2C scripts. Could you re-attach the audio files here so I can get a better idea of the issue? If you have scope captures that would be helpful as well. Also, what microphone are you using in the handsfree setup? 

For your issue of "shrill at higher volumes" and the echo, can you describe more what behavior this is? Do you want to implement a low pass filter for high pitched noise? Does the device have two microphones and two speakers, one set for handset and one for handsfree? Again, audio recording or scope capture of this echo would be helpful to figure out how to best help here.

The B0_P0_R84/83 register is the digital volume control for the ADC, which if increased, it would increase both the signal and the noise that the microphone picks up. 

B0_P0_R88 is the max PGA gain that the AGC will allow. Other important AGC parameters to consider would be the target level (B0_P0_R86) and noise threshold (B0_P0_R87).

Let me know more about the specifics of what you need help with and I can suggest settings to change.

Best,
Mir

Hi,

I am attaching the zip file that contains both the audio and the scripts. The explanation is provided below.This 1 KHz Sine tone recording , hope it would be helpful for you.Please refer capture diagram also.

In the Recordings_Phone folder, there are three folders, each containing three files:

   HFtoHF_Recording
       Audio_at_UserA_mic_recorded_HFtoHF_call.pcm: Recording from User A's mic before going to the network.
       Audio_received_at_UserB_HFtoHF_call.pcm: Recording received by User B from the network, made before playback through the speaker.

   HFtoHS_Recording
       Audio_at_UserA_mic_recorded_HFtoHS_call.pcm: Recording from User A's mic before going to the network.
       Audio_received_at_UserB_HFtoHS_call.pcm: Recording received by User B from the network, made before playback through the speaker.

   HStoHS_Recording
       Audio_at_UserA_mic_recorded_HStoHS_call.pcm: Recording from User A's mic before going to the network.
       Audio_received_at_UserB_HStoHS_call.pcm: Recording received by User B from the network, made before playback through the speaker.

Please a tool which can open .pcm audio files.

Handset Mode (HS): This mode usually features a microphone built into the handset of a VoIP phone (similar to a landline), placing the microphone closer to the speaker's mouth.

Hands-Free Mode(HF): This mode usually employs a speakerphone with a separate microphone from the handset, placing the microphone farther away from the user. As a result, it tends to capture more ambient sound.

Detail of the signal chain :

Analog Mic → Audio-codec (TLV32AIC3262) → SOC(AM5728)→→ → → Network → → → → Audio-codec (TLV32AIC3262) → SOC(AM5728) → Handset Speaker / Hands-free Speaker

Your Questions:

1. Could you re-attach the audio files here so I can get a better idea of the issue?                                                        :- Yes Attached and explanation as above
2. what microphone are you using in the hands-free setup?                                                                                      :- Mic Model- Sky CMEJ-0627-42-SP ,  Speaker Model- PUI Audio AS03608MR-4-R (If you have a better suggestion, we can implement that)
   
3. For your issue of "shrill at higher volumes" and the echo, can you describe more what behavior this is?                                                                                                                                                                            :- When one person is using the handset and another is on hands-free/speaker mode, the person on the handset is experiencing an echo.The shrill (Signal Clipping) sound is noticeable only at higher volumes in hands-free mode. Can we use DRC  (Dynamic range compression) ? if yes , Please suggest the configuration.
   
4. Do you want to implement a low pass filter for high pitched noise?                                                                      :- Yes if it can solve the issue
   
5. Does the device have two microphones and two speakers, one set for handset and one for hands-free?             :- Yes            
   Mic (Both HF and HS) : Sky CMEJ-0627-42-SP
   Speaker (HF) : PUI Audio AS03608MR-4-R
   Speaker (HS) : PUI Audio AR027150MR-2-R
   

TI_Reply_10102024.zip

Hi Akshi,

Thanks for your response. Could you change the format of the audio files to wav so I can open them? Also, can you attach scope shots of the noise you are experiencing, as well as the echo and the signal clipping sound? Let's make sure we are fully aware of the problem before I start recommending things.

And yes, you could use DRC to help reduce a loud sound, I would suggest looking at the section 2.5.4.3 of the applications reference guide for the part, it describes the DRC parameters and which registers control it. 

Let me know if you need more help with DRC, and please send over the audio files in a more universal format and scope captures of your issues. Thanks!

Best,
Mir

Hello Jeffres,

Thanks for your reply. I have attached a zip file with audio files in wav format. There are recording for sine wave input and multitone input in the zip file.

TI_Reply_301024.zip

The I2S waveform is also attached along with the readme are attached in below ZIP file.

I2S Signals_20241010.zip

Hi,

Sorry, I didn't have time to look into this today, I will respond tomorrow with my thoughts on tuning the AGC and the echo.

Best,
Mir

Hi,

Now that I am listening to the audio files, I am a little unsure of the issue. The record and playback seem to be at different sample rates so they are not clear what is the intended recording. Another thing to note is that DRC is intended for voice recordings, so you should use a sample of a voice recording as the input to the system to test it appropriately. 

To resolve your echo, you may just want to ramp down and then ramp up the volume when the mode is switched. 

If you can give some more explanation about the issue you are facing I can help more. Also, if you can paste in just the text that is your I2C commands, that will be easier to keep track of. Thanks.

Best,
Mir

Hi,

The recorded and playback audio are at different sampling rates because we are sampling the captured signal at 16khz in audiocodec and then in SIP handler we are using PCMU/PCMA (narrow band : 8khz) codec for encoding the media before sending over network.

In handsfree mode we are getting some continuous hissing noise along with the actual voice.

Hi,

Can you attach a schematic as well as scope shots of the hissing noise? I would like to test your script on a board here, to check the noise levels. Do you know what dB volumes you expect coming in to the device? What volumes are you getting in to the device? 

Best,
Mir

Hello Jeffres,

I have attached the Audio schematics below :

Audio_Schematics.pdf

It is not possible to get the scope shots of the noise. You can analyze the wav file( Audio_recived_at_UserB_HFtoHS.wav) for the noise captured.

We are getting around 50-60 db volume levels into the device. The input is human voice on a desk phone in handsfree mode.            

Hi,

So sorry, I meant to ask for a script as well. Do you have an I2C script that sets the registers of the device? Also, to be clear, you are using the TLV320AIC3212, not the AIC3262? 

-Mir

Hello,

We had already attached the i2c scripts in the earlier post : https://e2e.ti.com/support/audio-group/audio/f/audio-forum/1415034/tlv320aic3262-background-noise-observed-in-handset-mode

Click on the below link for the scripts for handsfree and handset mode :

Phone_mic_capture(2).zip

We are using TLV320AIC3262 and the wiring is same as shown in the schematics.

Hi Akshi,

Sounds good about the 3262. I wonder if you are able to move the speaker away from the microphone physically in handsfree mode. This may help. Another thing to consider is that at lower sample rates, the SNR will be worse as low frequency noise makes up more of the recorded spectrum. I'm not sure what noise you are referring to in these recordings. I compared "Audio_at_UserA_mic_recorded_HFtoHS_call" and "Audio_recived_at_UserB_HFtoHS_call" using Audacity, and overlaid them with their different sample rates. We can see in the spectrogram view that there are less of the harmonics of the 1k tone with the 8k sample rate recording. Are you hearing those round looking areas of more frequencies around the 1k line in the 8kHz recording? I wonder if that is caused by DRC. What happens if you disable the DRC with your script? 

Let me know about this.

Best,
Mir