audio module

The audio module provides the backbone for handling basic audio bookkeeping such as volume and signal routing. Actual sound is created by the engines, i.e. bl00mbox and media player at the moment. If you wish to add your own C-based sound engine to flow3r, please check out the framework we have set up for that in components/st3m/st3m_audio.*.

Jack Detection

audio.headset_is_connected() bool

Returns 1 if headphones with microphone were connected to the headphone jack at the last call of audio_update_jacksense.

audio.headphones_are_connected() bool

Returns 1 if headphones with or without microphone were connected to the headphone jack at the last call of audio_update_jacksense.

audio.line_in_is_connected() bool

Returns 1 if the line-in jack was connected at the last call of audio_update_jacksense.

Input Sources

Note

The onboard digital mic turns on an LED on the top board if it receives a clock signal which is considered a good proxy for its capability of reading data. Access to the onboard mic can be disabled entirely in the audio config menu.

The codec can receive data from the line in jack, the headset mic pin in the headphone jack, or the internal microphone. We distinguish between two use cases: 1) sending the signal to the audio engines to be mangled with, and 2) directly mix it to the output with a variable volume level. We provide two different APIs for each use case.

Sources may or may not be available; for line in and the headset mic they might simply not be plugged in, but also users can configure in the settings which inputs are available in the first place. To handle this uncertainity gracefully, instead of momentarily trying (and potentially failing) to set up a connection you set a desired target and the backend will attempt to connect to it continuously until the target is reset to none. The target states are not cleared when exiting applications, if you don’t intend to also configure the source for other applications please reset them to whatever state you found them in like so:

on_enter(self, vm):
    # save original target
    self.orig_engine_target_source = audio.input_engine_get_target_source()
    # switch to your preferred one
    audio.input_engine_set_target_source(audio.INPUT_SOURCE_AUTO)

on_exit(self):
    # restore original target
    audio.input_engine_set_source(self.orig_engine_target_source)

Since the codec can only send data from one source at a time. in case of a disagreement between the engine source and the thru source, the engine source wins and the through source is temporarily set to none. For thru to follow the engine source if available and not none use audio.input_thru_set_source(audio.INPUT_SOURCE_AUTO).

The available sources for both engine and thru are slightly different: The engine only looks for permissions and hardware state, while thru can not access the onboard mic if playback is happening through the speakers. This is set up to prevent accidential feedback loops. However the user can give permission to acces this mode in the user config.

audio.INPUT_SOURCE_NONE

No source, datastream suspended.

audio.INPUT_SOURCE_LINE_IN

Stream data from line in if available.

audio.INPUT_SOURCE_HEADSET_MIC

Stream data from headset mic if available and allowed.

audio.INPUT_SOURCE_ONBOARD_MIC

Stream data from onboard mic if allowed.

audio.INPUT_SOURCE_AUTO

Stream data from available input, INPUT_SOURCE_LINE_IN is preferred to INPUT_SOURCE_HEADSET_MIC is preferred to INPUT_SOURCE_ONBOARD_MIC.

For input_thru_set_source() only: matching input_engines_get_source() unless it is INPUT_SOURCE_NONE has highest preference, and INPUT_SOURCE_ONBOARD_MIC is never returned when speakers are on even if access is permitted.

audio.input_engines_set_source(source: int) int

Set up a continuous connection query for routing the given source to the input for the audio engines. Check for success with input_engines_get_source() and clean up by passing INPUT_SOURCE_NONE

audio.input_engines_get_target_source() int

Returns target source last set with i``nput_engines_set_source()``.

audio.input_engines_get_source() int

Returns source currently connected to the audio engines.

audio.input_engines_get_source_avail(source: int) bool

Returns true if it is currently possible to connect the audio engines to a given source. If given INPUT_SOURCE_AUTO returns true if any source can be connected to the engines.

audio.input_thru_set_source(source: int) int

Set up a continuous connection query for routing the given source to the output mixer of the codec. Check for success with input_thru_get_source() and clean up by passing INPUT_SOURCE_NONE

audio.input_thru_get_target_source() int

Returns target source last set with input_thru_set_source.

audio.input_thru_get_source() int

Returns the source currently mixed directly to output.

audio.input_get_source() int

Returns the source the codec is connected to at the moment.

audio.input_thru_get_source_avail(source: int) bool

Returns true if it is currently possible to a given source to thru. If given INPUT_SOURCE_AUTO returns true if any source can be connected to thru.

audio.input_thru_set_volume_dB(vol_dB: float)
audio.input_thru_get_volume_dB() float
audio.input_thru_set_mute(mute: bool)
audio.input_thru_get_mute() bool

Volume and mute control for input_thru. Please don’t use this as a replacement for terminating a connection, input_thru_set_source(audio.INPUT_SOURCE_NONE) instead!

audio.input_line_in_get_allowed(mute: bool)
audio.input_headset_mic_get_allowed(mute: bool)
audio.input_onboard_mic_get_allowed(mute: bool)
audio.input_onboard_mic_to_speaker_get_allowed(mute: bool)

Returns if the user has forbidden access to the resource.

OS development

Warning

These functions are not to be used in applications, but only by OS settings.

Many of these functions are available in three variants: headphone volume, speaker volume, and volume. If headphones_are_connected() returns 1 the “headphone” variant is chosen, else the “speaker” variant is chosen.

audio.headphones_detection_override(enable: bool, override_state: bool)

Set to ‘enable = True’ if the system should ignore jacksense and use override_state to determine whether headphones are connected (True) or not (False).

Use cases:

  • If a sleeve contact mic doesn’t pull the detection pin low enough the codec’s built in headphone detection might fail.

  • If the headset only has a mic connected but we wish to use the internal speaker anyway

Call with ‘enable = 0’ to revert to automatic detection.

audio.headphones_set_volume_dB(vol_dB: float) float
audio.speaker_set_volume_dB(vol_dB: float) float
audio.set_volume_dB(vol_dB: float) float

Attempts to set target volume for the headphone output/onboard speakers respectively, clamps/rounds if necessary and returns the actual volume. Absolute reference arbitrary. Does not unmute, use audio_{headphones_/speaker_/}set_mute as needed. Enters fake mute if requested volume is below the value set by audio_{headphones/speaker}_set_minimum_volume_user.

Note: This function uses a hardware PGA for the coarse value and software for the fine value. These two methods are as of yet not synced so that there may be a transient volume “hiccup”. “p1” badges only use software volume.

audio.headphones_adjust_volume_dB(vol_dB: float) float
audio.speaker_adjust_volume_dB(vol_dB: float) float
audio.adjust_volume_dB(vol_dB: float) float

Like the audio_{headphones_/speaker_/}set_volume family but changes relative to last volume value.

audio.headphones_get_volume_dB() float
audio.speaker_get_volume_dB() float
audio.get_volume_dB() float

Returns volume as set with audio_{headphones/speaker}_set_volume_dB.

audio.headphones_get_mute() int
audio.speaker_get_mute() int
audio.get_mute() int

Returns 1 if channel is muted, 0 if channel is unmuted.

audio.headphones_set_mute(mute: int)
audio.speaker_set_mute(mute: int)
audio.set_mute(mute: int)

Mutes (mute = 1) or unmutes (mute = 0) the specified channel.

Note: Even if a channel is unmuted it might not play sound depending on the return value of audio_headphone_are_connected. There is no override for this (see HEADPHONE PORT POLICY below).

audio.headphones_set_minimum_volume_dB(vol_dB: float) float
audio.speaker_set_minimum_volume_dB(vol_dB: float) float
audio.headphones_set_maximum_volume_dB(vol_dB: float) float
audio.speaker_set_maximum_volume_dB(vol_dB: float) float

Set the minimum and maximum allowed volume levels for speakers and headphones respectively. Clamps with hardware limitations. Maximum clamps below the minimum value, minimum clamps above the maximum. Returns clamped value.

audio.headphones_get_minimum_volume_dB() float
audio.speaker_get_minimum_volume_dB() float
audio.headphones_get_maximum_volume_dB() float
audio.speaker_get_maximum_volume_dB() float

Returns the minimum and maximum allowed volume levels for speakers and headphones respectively. Change with audio_{headphones/speaker}_set_{minimum/maximum}_volume_dB.

audio.headphones_get_volume_relative() float
audio.speaker_get_volume_relative() float
audio.get_volume_relative() float

Syntactic sugar for drawing UI: Returns channel volume in a 0..1 range, scaled into a 0.01..1 range according to the values set with audio_{headphones_/speaker_/}set_{maximum/minimum}_volume_ and 0 if in a fake mute condition.

audio.headset_mic_set_gain_dB(gain_dB: float)
audio.headset_mic_get_gain_dB() float
audio.onboard_mic_set_gain_dB(gain_dB: float)
audio.onboard_mic_get_gain_dB() float
audio.line_in_set_gain_dB(gain_dB: float)
audio.line_in_get_gain_dB() float

Set and get gain for the respective input channels.

audio.codec_i2c_write(reg: int, data: int)

Write audio codec register. Obviously very unsafe. Do not use in applications that you distribute to users. This can fry your speakers with DC

audio._codec_configure_dynamic_range_control(enable: bool, attack: int, release: int, make_up_gain_dB: int, comp_ratio: int, comp_threshold_dB: int, exp_ratio: int, exp_threshold_dB: int)

Configures dynamic range control (compressor/expander) in the max98091 codec.

Temporary API for compressor experimentation, to be replaced later.

See page 134 of the datasheet for values:

https://www.analog.com/media/en/technical-documentation/data-sheets/MAX98091.pdf

exp_threshold_dB is expressed in dB from 35 to 66, not in raw values from 0 to 31.

Example:

audio._codec_configure_dynamic_range_control(True, 2, 3, 0, 4, 10, 0, 0)

Will result in enabling the compressor with attack 2 (1.25ms), release 3 (1s), no make up gain, compression ratio 4 (INF:1), compression threshold 10dB, and expander disabled (ratio 1:1)

Headphone port policy

Under normal circumstances it is an important feature to have a reliable speaker mute when plugging in headphones. However, since the headphone port on the badge can also be used for badge link, there are legimate cases where it is desirable to have the speakers unmuted while a cable is plugged into the jack.

As a person who plugs in the headphones on the tram, doesn’t put them on, turns on music to check if it’s not accidentially playing on speakers and then finally puts on headphones (temporarily, of course, intermittent checks if the speakers didn’t magically turn on are scheduled according to our general anxiety level) we wish to make it difficult to accidentially have sound coming from the speakers.

Our proposed logic is as follows (excluding boot conditions):

  1. Badge link TX cannot be enabled for any of the headphone jack pins without a cable detected in the jack. This is to protect users from plugging in headphones while badge link is active and receiving a short but potentially very loud burst of digital data before the software can react to the state change.

  2. If the software detects that the headphone jack has changed from unplugged to plugged it always turns off speakers, no exceptions.

  3. If a user wishes to TX on headphone badge link, they must confirm a warning that having headphones plugged in may potentially cause hearing damage every time.

  4. If a user wishes to RX or TX on headphone badge link while playing sound on the onboard speakers, they must confirm a warning every time.

We understand that these means seem extreme, but we find them to be a sensible default configuration to make sure people can safely operate the device without needing to refer to a manual.

(TX here means any state that is not constantly ~GND with whatever impedance. While there are current limiting resistors (value TBD at the time of writing, but presumably 100R-470R) in series with the GPIOs, they still can generate quite some volume with standard 40Ohm-ish headphones. Ideally the analog switch will never switch to the GPIOs without a cable plugged in.)