ConHiveConHive Docs
ModulesAudio Management

Audio Management workflow nodes

Workflow builder nodes of the Audio Management module.

These nodes appear in the workflow builder when the Audio Management module is available.

Play Audio

This node plays an audio file on the device's default audio output. It is useful for playing back recorded audio, sound effects, or music files as part of an application or system functionality. The node requires the file path of the audio to be played, allowing for dynamic selection of audio content. Once triggered, the node will play the audio file through the device's audio output system, providing audible feedback or entertainment to users.

Inputs

NameDescriptionTypeRequired
FilepathThe file path of the audio file to be played. This specifies the location of the audio content to be played back.textYes

Test Speaker

This node plays a test sound on the device's default speaker, helping to verify that the audio output system is functioning correctly. It can be used in troubleshooting or as part of system diagnostics to ensure that the sound hardware is properly configured and working. The test is triggered by a boolean input, allowing it to be easily integrated into automated workflows for audio testing.

Inputs

NameDescriptionTypeRequired
StartTestA boolean trigger to start the speaker test. When true, the node will emit a test sound through the default audio output.yes/noYes

Settings

FieldDescriptionTypeRequiredAllowed valuesDefault
VolumeThe volume level of the speaker test.numberNo0 to 10050
LoopsThe number of loops for the speaker test.numberNo1 to 102
ChannelsThe number of audio channels for the test.numberNo1 to 22
Card nameThe name of the audio card to use for the test.textNo0 to 250 characters
Device nameThe hardware device identifier for the speaker test.textNo0 to 250 characters

Device List

Queries the operating system for all available audio input and output devices and emits their metadata (device name, index, hardware ID, supported formats) as a single structured payload on the Devices output. This node is primarily a diagnostic and setup tool — run it once to discover which device names to enter in DeviceCapture or AudioPlayback configurations. Because it is marked is_start and custom='auto', it executes at pipeline start automatically. Example: run a pipeline consisting of just DeviceList → ConsoleEmitter to print all available audio devices to the terminal; use the printed names to configure your capture nodes.

Outputs

NameDescriptionType
DevicesDiscovered device informationgroup of fields (list)

Settings

FieldDescriptionTypeRequiredAllowed valuesDefault
DirectionFilter: 'input', 'output', or '' for alltextNoat most 10 characters

Monitoring Recorder

Long-term audio recording node. Continuously writes incoming SignalBlocks into rolling files organised by date/hour folders. Output format is selectable: FLAC (lossless compressed, 16/24-bit) for archival or WAV (uncompressed PCM, 16/24/32-bit) for ML training pipelines. Each segment ships with a sidecar JSON containing capture metadata (sample rate, channels, codec, bit depth, device, agent version, muted intervals). Optionally mutes detected speech regions with pink noise based on Silero VAD for privacy. Designed for long-term wildlife monitoring (e.g. bird research) where uninterrupted operation matters more than 100% capture coverage; IO errors are logged but never propagate, so the surrounding flow keeps running. Built-in retention by file age and total disk usage.

Inputs

NameDescriptionTypeRequired
SignalInput audio signal block — interleaved multi-channel samples plus sample rate / channel count metadata.group of fieldsYes

Outputs

NameDescriptionType
SignalPass-through of the unmodified input block. Muting only affects the on-disk copy, downstream stages still see the original samples.group of fields (list)
StatusEventLive recorder status (side-channel via OnOutput): active segment path, sample rate, channels and format.group of fields (list)

Settings

FieldDescriptionTypeRequiredAllowed valuesDefault
Base dirBase directory under which date/hour folders and recordings are created.textYes1 to 512 characters/var/lib/conhive/recordings
Folder patternGo time-layout determining the sub-folder hierarchy under base_dir. Example: '2006-01-02/15' yields date/hour buckets.textNo1 to 64 characters2006-01-02/15
Filename prefixPrefix prepended to the segment timestamp in the recording filename.textNo1 to 32 charactersrec
Max duration sMaximum duration of a single segment in seconds. Recording rotates to a fresh file when this is reached.numberNo10 to 3600300
Bit depthSample bit depth for flac/wav. FLAC supports 16 or 24; WAV supports 16, 24 or 32. 32-bit is rejected for FLAC (the encoder cannot produce valid 32-bit streams). Ignored for the wav_float32 format, which is always 32-bit float.numberNo16 to 3216
FormatOutput container format. flac = lossless compressed (16/24-bit); wav = uncompressed integer PCM (16/24/32-bit); wav_float32 = uncompressed 32-bit IEEE float, an exact passthrough of the audio bus (zero requantisation) that numpy/torchaudio load natively — the preferred choice for ML training. FLAC is best for long-term archival.choiceNoflac; wav; wav_float32flac
Max age hoursRetention by file age in hours. 0 disables age-based pruning.numberNoat least 00
Max disk bytesRetention by aggregate recording size in bytes. 0 disables size-based pruning.numberNoat least 00
Mute speechWhen true, runs Silero VAD on each block and overwrites detected speech regions with pink noise across all channels before writing to disk.yes/noNofalse
Speech thresholdSilero VAD probability threshold above which a frame is considered speech.numberNo0 to 10.5
Write metadataWhen true, writes a .json sidecar next to each finalised FLAC segment with capture metadata and muted intervals.yes/noNotrue