HFP-AG example not working as intended (IDFGH-6114)

[MarcosHWconfig]

Environment

• Development Kit: ESP32-LyraT v4.3
• Module or chip used: ESP32-WROVER-E
• IDF version: v4.3.1-245-gba15ac8634
• Build System: idf.py
• Compiler version: xtensa-esp32-elf-gcc --version
• Operating System: Windows
• (Windows only) environment type: ESP Command Prompt
• Using an IDE?: No
• Power Supply: USB

Problem Description

I flashed the esp-idf\examples\bluetooth\bluedroid\classic_bt\hfp_ag example in my LyraT after modifying the following parameters from the menuconfig:

• BR/EDR Sync(SCO/eSCO) max connections --> 1
• BR/EDR Sync(SCO/eSCO) default data path --> HCI
• audio(SCO) data path --> HCI
• Wide band speech --> ON

From the example code I see that the AG sends a sinusoid to the connected client. I used several pairs of bluetooth headphones (fully functional) as clients but the only thing I hear is noise, definitely not a sinusoid. I have recorded the received signal, here is the plot:

Expected Behavior

Hear the sinusoid correctly from the client side.

Actual Behavior

I hear a strange noise from the client side, not a sinusoid.

Steps to reproduce

1. Get the code esp-idf\examples\bluetooth\bluedroid\classic_bt\hfp_ag
2. Change BR/EDR Sync(SCO/eSCO) max connections --> 1, BR/EDR Sync(SCO/eSCO) default data path --> HCI, audio(SCO) data path --> HCI, Wide band speech --> ON
3. Edit app_hf_msg_set.c and set the MAC address of the client
4. Open serial console, type con and after connection type cona
5. Hear the noise from the client

[skateskate]

I am testing the same configuration as @MarcosHWconfig and I also have the same problem with the sinusoid.
Can someone look into it? Maybe @Wth-Esp can help us?

[ghost]

hi all,

my colleague is handling this issue, could you @BetterJincheng please provide a patch here?

Thanks!

[BetterJincheng]

patch.zip

Hi all,

This is the patch for this issue towards IDF_V4.3.
Try to use it by :
unzip patch.zip
git apply [patch_name]

Turn to us if any problems.

[skateskate]

Hello @BetterJincheng and @Wth-Esp, thanks for the patch.
On my system the sinusoid is now ok, but I can still hear some spikes now and then (let's say 3-4 spikes every 10 seconds). Is it normal?
If you don't mind I have another question: what is the purpose of the timer that runs out every 4000 microseconds (PCM_GENERATOR_TICK_US)? If I understand correctly, it is used to release the semaphore to the function that generates frame_data_num samples of the sinusoid, but why 4000 microseconds precisely? Could the spikes be related to this?

[BetterJincheng]

Hi @skateskate,

Codec (code & decode) for HFP is mSBC with specific parameters. It transfers 240 Bytes every 7.5 ms. So in the demo we need to create 240 Bytes every 7.5ms. PCM_GENERATOR_TICK_US is just the periodic duration we create the input audio data in the purpose of simulating a role of microphone. You can set it to any value as you want.