wakenet hard to trigger (AIS-2193)

[2832544382]

Checklist

• Checked the issue tracker for similar issues to ensure this is not a duplicate.
• Provided a clear description of your suggestion.
• Included any relevant context or examples.

Issue or Suggestion Description

The example i use is COZE_WS_APP, I tried the continuous mode nad it's work, but when I try to use the wake up mode, the wake word is hard to trigger, here's the logs and the menuconfig setting.

I (9640) ESP_GMF_POOL: Registered items on pool:0x3c314c00, audio_manager_init-229
I (9642) ESP_GMF_POOL: IO, Item:0x3c314e98, H:0x3c314df8, TAG:io_codec_dev
I (9649) ESP_GMF_POOL: IO, Item:0x3c314f48, H:0x3c314ea8, TAG:io_codec_dev
I (9655) ESP_GMF_POOL: IO, Item:0x3c315008, H:0x3c314f58, TAG:io_file
I (9662) ESP_GMF_POOL: IO, Item:0x3c3150c8, H:0x3c315018, TAG:io_file
I (9668) ESP_GMF_POOL: IO, Item:0x3c3151a8, H:0x3c3150d8, TAG:io_http
I (9674) ESP_GMF_POOL: IO, Item:0x3c315260, H:0x3c3151b8, TAG:io_embed_flash
I (9681) ESP_GMF_POOL: EL, Item:0x3c315364, H:0x3c315270, TAG:aud_enc
I (9687) ESP_GMF_POOL: EL, Item:0x3c315480, H:0x3c315374, TAG:aud_dec
I (9693) ESP_GMF_POOL: EL, Item:0x3c31564c, H:0x3c315490, TAG:aud_alc
I (9699) ESP_GMF_POOL: EL, Item:0x3c31572c, H:0x3c31565c, TAG:aud_ch_cvt
I (9706) ESP_GMF_POOL: EL, Item:0x3c315808, H:0x3c31573c, TAG:aud_bit_cvt
I (9712) ESP_GMF_POOL: EL, Item:0x3c3158ec, H:0x3c315818, TAG:aud_rate_cvt
I (9720) AUDIO_PROCESSOR: Audio manager initialized successfully
I (9724) AUDIO_PROCESSOR: Opening audio prompt...
W (9729) AUD_SDEC_REG: Overwrote ES decoder 542523735
W (9734) AUD_SDEC_REG: Overwrote ES decoder 1094792269
W (9739) AUD_SDEC_REG: Overwrote ES decoder 1398026829
I (9743) ASP_POOL: Dest rate:16000
I (9746) ASP_POOL: Dest channels:2
I (9750) ASP_POOL: Dest bits:32
I (9753) ESP_GMF_TASK: Waiting to run... [tsk:TSK_0x3fcda3ac-0x3fcda3ac, wk:0x0, run:0]
I (9754) AUDIO_PROCESSOR: Audio prompt opened successfully
I (9765) AUDIO_PROCESSOR: Opening audio recorder...
I (9770) MODEL_LOADER: The storage free size is 21504 KB
I (9775) MODEL_LOADER: The partition size is 5500 KB
I (9780) MODEL_LOADER: Successfully load srmodels
I (9784) AFE_CONFIG: Set WakeNet Model: wn9s_nihaoxiaozhi
I (9789) AFE_CONFIG: Set Second WakeNet Model: wn9s_nihaoxiaozhi

/********** General AFE (Audio Front End) Parameter **********/
pcm_config.total_ch_num: 2
pcm_config.mic_num: 1: [ ch0 ]
pcm_config.ref_num: 1: [ ch1 ]
pcm_config.sample_rate: 16000
afe_type: VC
afe_mode: HIGH PERF
afe_perferred_core: 0
afe_perferred_priority: 5
afe_ringbuf_size: 50
memory_alloc_mode: 3
afe_linear_gain: 1.0
debug_init: false
fixed_first_channel: false

/********** AEC (Acoustic Echo Cancellation) **********/
aec_init: true
aec mode: VOIP_HIGH_PERF
aec_filter_length: 4

/********** SE (Speech Enhancement, Microphone Array Processing) **********/
se_init: false, model: BSS

/********** NS (Noise Suppression) **********/
ns_init: true
ns model name: WEBRTC

/********** VAD (Voice Activity Detection) **********/
vad_init: true
vad_mode: 3
vad_model_name: NULL
vad_min_speech_ms: 64
vad_min_noise_ms: 1000
vad_delay_ms: 128
vad_mute_playback: false
vad_enable_channel_trigger: false

/********** WakeNet (Wake Word Engine) **********/
wakenet_init: true
wakenet_model_name: wn9s_nihaoxiaozhi
wakenet_model_name_2: wn9_hilexin
wakenet_mode: 0

/********** AGC (Automatic Gain Control) **********/
agc_init: false
agc_mode: WEBRTC
agc_compression_gain_db: 9
agc_target_level_dbfs: 3

/**************************************************/
MC Quantized wakenet9s: wakenet9s_tts2h8v2_你好小智_3_0.630_0.635, tigger:v4, mode:0, p:0, (Aug 11 2025 15:20:50)
MC Quantized wakenet9: wakenet9_v1h24_嗨，乐鑫_3_0.608_0.615, tigger:v4, mode:0, p:0, (Aug 11 2025 15:20:50)
I (9980) AFE: AFE Version: (1MIC_V250121)
I (9983) AFE: Input PCM Config: total 2 channels(1 microphone, 1 playback), sample rate:16000
I (9991) AFE: AFE Pipeline: [input] -> |AEC(VOIP_HIGH_PERF)| -> ilexin)| -> |NS(WebRTC)| -> |VAD(WebRTC)| -> |WakeNet(wn9s_nihaoxiaozhi,wn9_hilexin)| -> [output]
I (10005) AFE_MANAGER: Feed task, ch 2, chunk 256, buf size 1024
I (10005) GMF_AFE: Create AFE, ai_afe-0x3c351264
I (10015) AUDIO_PROCESSOR: AFE created & registered: 0x3c351264
I (10021) AUDIO_PROCESSOR: AFE created & registered: 0x3c351264
I (10026) AUDIO_PROCESSOR: WakeNet init flag = 1
I (10031) GMF_AFE: Create AFE, ai_afe-0x3c351340
I (10035) GMF_AFE: New an object,ai_afe-0x3c351340
I (10040) AUDIO_PROCESSOR: Recorder pipeline created with registered element pool
W (10047) AUDIO_PROCESSOR: Binding pipeline ai_afe (0x3c351340) with event callback
W (10054) ESP_GMF_PIPELINE: There is no thread for add jobs, pipe:0x3c3749a0, tsk:0x0, [el:ai_afe-0x3c351340]
I (10064) ESP_GMF_TASK: Waiting to run... [tsk:audio_recorder_task-0x3fcdc8bc, wk:0x0, run:0]
I (10072) NEW_DATA_BUS: New ringbuffer:0x3c376400, num:2, item_cnt:1024, db:0x3c376428
I (10080) NEW_DATA_BUS: New ringbuffer:0x3c376480, num:1, item_cnt:6144, db:0x3c3764a8
I (10087) AFE_MANAGER: AFE manager suspend 1
I (10091) AFE_MANAGER: AFE manager suspend 0
I (10095) ESP_GMF_TASK: One times job is complete, del[wk:0x3c376370, ctx:0x3c351340, label:ai_afe_open]
I (10073) AUDIO_PROCESSOR: PIPE=0x3c3749a0 TASK=0x3fcdc8bc
I (10110) AUDIO_PROCESSOR: AFE actually used by pipeline: 0x3c351340
I (10116) AUDIO_PROCESSOR: Audio recorder opened successfully
I (10121) AUDIO_PROCESSOR: Opening audio playback...
I (10126) NEW_DATA_BUS: New pbuf, num:5, item_cnt:1, db:0x3c3c5a10
W (10132) ESP_GMF_PIPELINE: There is no thread for add jobs, pipe:0x3c3c5a68, tsk:0x0, [el:aud_dec-0x3c3c5aac]
I (10142) ESP_GMF_TASK: Waiting to run... [tsk:audio_playback_task-0x3fcddc00, wk:0x0, run:0]
I (10150) ESP_GMF_TASK: Waiting to run... [tsk:audio_playback_task-0x3fcddc00, wk:0x3c3e2950, run:0]
I (10143) AUDIO_PROCESSOR: Audio playback opened successfully
I (10164) AUDIO_PROCESSOR: Starting audio playback...
I (10106) ESP_GMF_PORT: ACQ IN, new self payload:0x3c376370, port:0x3c376330, el:0x3c351340-ai_afe
W (10170) AUD_SDEC: Not find default parser for 1398100047
I (10183) ESP_GMF_TASK: One times job is complete, del[wk:0x3c3e2950, ctx:0x3c3c5aac, label:aud_dec_open]
I (10192) ESP_GMF_PORT: ACQ IN, new self payload:0x3c3e2950, port:0x3c3d88cc, el:0x3c3c5aac-aud_dec
I (10171) AUDIO_PROCESSOR: Audio playback started successfully

[sun-xiangyu]

srmodels I (9784) AFE_CONFIG: Set WakeNet Model: wn9s_nihaoxiaozhi I (9789) AFE_CONFIG: Set Second WakeNet Model: wn9s_nihaoxiaozhi

Please select wn9_nihaoxiaozhi model.
wn9s_nihaoxiaozhi is a wakenet9s model, which is designed for esp32c5.

[2832544382]

I have uploaded a video to record the problem, and I switched to wn9_nihaoxiaozhi model

https://youtube.com/shorts/EC8nifZV5W4?feature=share

[sun-xiangyu]

According to your logs, the esp-sr configuration appears to be correct. The video shows that you completely fail to wake the device. I suggest you first export the audio data and check it.