ESP-ADF (Espressif's Audio Development Framework) works on top of ESP-IDF, the commonly-used base SDK for ESP32. ESP-ADF provides maximum flexibility to application developers who want to use it as a complete solution supporting a range of use cases from provisioning to OTA, or use it as a platform for building fully-customized applications. 

ESP-ADF includes a rich set of features, such as codecs, source and sink streams, pipelining support, different services and controls, and even a wake-word engine.

Espressif's Audio Development Framework: 

• Supports popular audio formats: MP3, AAC, WAV, OGG, AMR, TS, OPUS, SPEEX, etc.
• Supports the creation of sound effects with tools such as: EQ, Mixer and Resample.
• Plays music from sources such as: HTTP, HLS (HTTP Live), SD card, Bluetooth A2DP/HFP.
• Integrates Media services such as: DLNA, Airplay, WeChat, Internet radio.
• Supports voice recognition and integration with online services: Alexa, DuerOS, Turing, IFLYTEK, TmallGenie, RooBo, etc. 

Potential applications deploying ESP-ADF include smart speakers, voice-activated walkie-talkies, broadcasters and other audio-enabled solutions, such as connected story-telling toys and point-reading pens. 

ESP-ADF is supported on Espressif’s LyraT and LyraTD-MSC development boards. The documentation and design of these boards are open-source, allowing users to easily create their own applications based on ESP-ADF. 

Additionally, Espressif’s ESP32-WROVER module, which has an integrated SPIRAM, can act as a powerful, feature-rich, yet cost-effective, platform for audio-application hardware. 

The entire ESP-ADF repository can be accessed here, while the diagram below encapsulates everything that ESP-ADF offers.