Dropouts

[schuh8]

I have built an internet radio based on the example code but on a TTGO T-display board. While it does seem to work I get a lot of dropouts with the serial message "stream to slow, dropouts possible". It would seem that too fast a stream, not too slow a stream would have dropout problems..The station is the BBC world service which works fine on another homebuilt IR based on the Heltec WIFI Kit 32 board with a VS1053 doing the decoding. Obviously the internet connection is fine. So is this a problem with the TTGO board or a basic fault of the Audio.h library? Thanks for any help you can offer.....

[jlarnal]

It happens to me too, on this very stream : https://strm112.1.fm/brazilianbirds_mobile_mp3
Even preparing the Audio instance with setConnectionTimeout() and ridiculous values won't help, and I have my idea why...

Two important things to notice about this stream :

1. It originates from a secure socket, obviously.
2. It's a 320Kbps MP3 (ouch) and plays out at 44100KHz (ouch, bis).

Other than that, the Audio::loop() method is fired each 5ms (approx) and the PSRAM buffer is 2MB (overkill ?)
But still, I wager both cited impediments are making the buffering and decoding heavy on computations, hence the dropouts.
I would bet a 192Kbps stream with 32KHz would sound exactly the same (those are birds songs playing) and limit buffer underruns -sigh- 😓.

Am I missing something here ?

@schuh8 , how "heavy" is your stream to decode compared to the one I gave the URL to ?
Is the stream SSL-encrypted ?

[jlarnal]

@schreibfaul1
Quick though ... would it be possible to force the helix decoder's most used methods on IRAM ?

https://docs.espressif.com/projects/esp-idf/en/latest/esp32s2/api-guides/memory-types.html

[schreibfaul1]

I can't say how effective the IRAM usage is or if there are negative side effects.
https://strm112.1.fm/brazilianbirds_mobile_mp3 returns 16KB blocks after decryption.
You can see this by looking at availableBytes = _client-available();. There are often more than 400ms between the blocks. This empties the audio buffer faster than it fills it up. With the unencrypted stream http://strm112.1.fm/brazilianbirds_mobile_mp3 the data comes continuously.

[jlarnal]

Ooooh... So it's a server latency problem then (combined with an overuse of encryption) ? 😯

Thanks @schreibfaul1 😉

[schuh8]

Little confused on that one. I have another homemade Internet Radio based on a ESP32 with a VS1053 and have no problems with dropouts on the literally hundreds of stations that I've tried. Maybe the VS1053 has some sort of built in buffer ?? However not using a VS1053 makes for a more compact radio with lesser current draw.

[chess-levin]

I'm joining this thread, because of having the same dropouts problems as you guys :-(

The dropouts happening with 64kbit, 128kbit, 192kbit and different sample rates starting from 24000.

My sketch is based on @schreibfaul1 ESP32_audioIS example sketch. Audiooutput goes via I2S -> PCM5102 -> Headphone

While I'm testing there is no LCD Display code active.

Here are some log outputs:

CONFIGURATION: https://docs.platformio.org/page/boards/espressif32/esp32dev.html
PLATFORM: Espressif 32 (6.4.0) > Espressif ESP32 Dev Module
HARDWARE: ESP32 240MHz, 320KB RAM, 4MB Flash
DEBUG: Current (cmsis-dap) External (cmsis-dap, esp-bridge, esp-prog, iot-bus-jtag, jlink, minimodule, olimex-arm-usb-ocd, olimex-arm-usb-ocd-h, olimex-arm-usb-tiny-h, olimex-jtag-tiny, tumpa)
PACKAGES:
 - framework-arduinoespressif32 @ 3.20011.230801 (2.0.11)
 - tool-esptoolpy @ 1.40501.0 (4.5.1)
 - tool-mkfatfs @ 2.0.1 
 - tool-mklittlefs @ 1.203.210628 (2.3)
 - tool-mkspiffs @ 2.230.0 (2.30)
 - toolchain-xtensa-esp32 @ 8.4.0+2021r2-patch5
LDF: Library Dependency Finder -> https://bit.ly/configure-pio-ldf
LDF Modes: Finder ~ chain, Compatibility ~ soft
Found 37 compatible libraries
Scanning dependencies...
Dependency Graph
|-- ESP32-audioI2S-master @ 2.0.0+sha.4e95bba
|-- Ai Esp32 Rotary Encoder @ 1.6.0
|-- LiquidCrystal_I2C_ESP32 @ 1.1.6
|-- ArduinoLog @ 1.1.1
|-- WiFi @ 2.0.0
Building in debug mode
Compiling .pio\build\esp32dev\src\main.cpp.o
Linking .pio\build\esp32dev\firmware.elf
Retrieving maximum program size .pio\build\esp32dev\firmware.elf
Checking size .pio\build\esp32dev\firmware.elf
Advanced Memory Usage is available via "PlatformIO Home > Project Inspect"
RAM:   [==        ]  15.3% (used 50188 bytes from 327680 bytes)
Flash: [========= ]  92.5% (used 1211877 bytes from 1310720 bytes)
Building .pio\build\esp32dev\firmware.bin
esptool.py v4.5.1
Creating esp32 image...
Merged 25 ELF sections
Successfully created esp32 image.
Configuring upload protocol...
AVAILABLE: cmsis-dap, esp-bridge, esp-prog, espota, esptool, iot-bus-jtag, jlink, minimodule, olimex-arm-usb-ocd, olimex-arm-usb-ocd-h, olimex-arm-usb-tiny-h, olimex-jtag-tiny, tumpa
CURRENT: upload_protocol = esptool
Looking for upload port...
Auto-detected: COM8
Uploading .pio\build\esp32dev\firmware.bin
esptool.py v4.5.1
Serial port COM8
Connecting.....
Chip is ESP32-D0WD-V3 (revision v3.1)
Features: WiFi, BT, Dual Core, 240MHz, VRef calibration in efuse, Coding Scheme None
Crystal is 40MHz

I: Chip model ESP32-D0WD-V3 (rev 3), #cores 2, CPU Freq 240 MHz
I: FlashChipSize 4096 KB, FlashChipSpeed 40000000

Example 1: A mid quality stream:

info        Connect to new host: "http://kcrw.streamguys1.com/kcrw_192k_mp3_on_air"
info        buffers freed, free Heap: 146996 bytes
info        Connection has been established in 470 ms, free Heap: 146392 bytes
bitrate     192000
icyurl      www.kcrw.com
station     KCRW Live
info        MP3Decoder has been initialized, free Heap: 121824 bytes , free stack 2088 DWORDs
lasthost    http://kcrw.streamguys1.com/kcrw_192k_mp3_on_air
info        StreamTitle='IN THIS WORLD (feat. Ryuichi Sakamoto & Hikari Mitsushima)-MONDO GROSSO-BIG WORLD'
streamtitle IN THIS WORLD (feat. Ryuichi Sakamoto & Hikari Mitsushima)-MONDO GROSSO-BIG WORLD
info        stream ready
info        syncword found at pos 0
info        MP3 decode error -2 : MAINDATA_UNDERFLOW
info        syncword found at pos 62
info        syncword found at pos 0
info        MP3 decode error -6 : INVALID_FRAMEHEADER
info        syncword found at pos 443
info        syncword found at pos 0
info        MP3 decode error -9 : INVALID_HUFFCODES
info        syncword found at pos 118
info        syncword found at pos 0
info        MP3 decode error -9 : INVALID_HUFFCODES
info        syncword found at pos 625
info        syncword found at pos 0
info        Channels: 2
info        SampleRate: 44100
info        BitsPerSample: 16
info        BitRate: 192000
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        StreamTitle='This Version of You (feat. Julianna Barwick) [Joseph Ray Remix]-ODESZA-Buddha Bar by Christos Fourkis & Ravin'
streamtitle This Version of You (feat. Julianna Barwick) [Joseph Ray Remix]-ODESZA-Buddha Bar by Christos Fourkis & Ravin
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible

Example 2: A low quality stream:

info        Connect to new host: "http://rnrw.cast.addradio.de/rnrw-0182/dein90er/low/stream.mp3"
info        buffers freed, free Heap: 147384 bytes
info        Connection has been established in 75 ms, free Heap: 146796 bytes
info        redirect to new host "http://rnrw--di--rnrw-ais-01--03--cdn.cast.addradio.de/rnrw/dein90er/low/stream.mp3?_art=dj0yJmlwPTc3LjIyLjEzOS4yMTAmaWQ9aWNzY3hsLTN4eDdodHlrYiZ0PTE3MDQ3MjEzNDAmcz03ODY2ZjI5YyM3NzVmNzhlMTYxM2QxOWJlOWI4ZWRhYzQ1NGM0YmZjMg"
info        Connect to new host: "http://rnrw--di--rnrw-ais-01--03--cdn.cast.addradio.de/rnrw/dein90er/low/stream.mp3?_art=dj0yJmlwPTc3LjIyLjEzOS4yMTAmaWQ9aWNzY3hsLTN4eDdodHlrYiZ0PTE3MDQ3MjEzNDAmcz03ODY2ZjI5YyM3NzVmNzhlMTYxM2QxOWJlOWI4ZWRhYzQ1NGM0YmZjMg"
info        buffers freed, free Heap: 146336 bytes
info        Connection has been established in 87 ms, free Heap: 145900 bytes
bitrate     64000
icyurl
info        MP3Decoder has been initialized, free Heap: 122052 bytes , free stack 5560 DWORDs
lasthost    http://rnrw--di--rnrw-ais-01--03--cdn.cast.addradio.de/rnrw/dein90er/low/stream.mp3?_art=dj0yJmlwPTc3LjIyLjEzOS4yMTAmaWQ9aWNzY3hsLTN4eDdodHlrYiZ0PTE3MDQ3MjEzNDAmcz03ODY2ZjI5YyM3NzVmNzhlMTYxM2QxOWJlOWI4ZWRhYzQ1NGM0YmZjMg
info        stream ready
info        syncword found at pos 0
info        MP3 decode error -2 : MAINDATA_UNDERFLOW
info        syncword found at pos 72
info        syncword found at pos 0
info        MP3 decode error -6 : INVALID_FRAMEHEADER
info        syncword found at pos 118
info        syncword found at pos 0
info        Channels: 2
info        SampleRate: 24000
info        BitsPerSample: 16
info        BitRate: 64000
info        StreamTitle='MEREDITH BROOKS - BITCH'
streamtitle MEREDITH BROOKS - BITCH
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        StreamTitle='ANN LEE - 2 TIMES'
streamtitle ANN LEE - 2 TIMES
info        StreamTitle='THE CURE - FRIDAY I'M IN LOVE'
streamtitle THE CURE - FRIDAY I'M IN LOVE
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible
info        slow stream, dropouts are possible

• I've seen there is another example called examples/Better_WiFi_throughput, but I don't understand what exactly to do with it: Do I have to make the described TCP configurations within the idf IOT Development Framework only once and afterwards I can go back to PlatformIO to develop & upload my sketches?
• Should I switch from my WROOM to a WROVER module?
• Any help on the dropout issue appreciated!