Fix pitch issue (only) with dynamic tracks and sync #12515
Conversation
This prevents a sudden pitch change in the leader deck.
…e postPorcess() This fixes the issue of changing pitch in two clones playing in sync
This is required for a reliable sync feature with changing tempo tracks. All other scalers have already double precision.
…fter track is loaded
…g the bpm after track load.
| @@ -25,6 +25,7 @@ class EngineDeck : public EngineChannel, public AudioDestination { | |||
|
|
|||
| void process(CSAMPLE* pOutput, const int iBufferSize) override; | |||
| void collectFeatures(GroupFeatureState* pGroupFeatures) const override; | |||
| void postProcessLocalBpm() override; | |||
There was a problem hiding this comment.
can you add comments to these definitions, like:
// postProcessLocalBpm is called on all decks to update the local bpm after processing is done, because bpm can vary depending on track position which is only known after processing.
// postProcess is called on all decks after postProcessLocalBpm to update beat distances, sync modes, and other values that are only known after all other processing is done.
src/engine/enginebuffer.cpp
Outdated
| @@ -1332,6 +1339,13 @@ void EngineBuffer::processSeek(bool paused) { | |||
| m_queuedSeek.setValue(kNoQueuedSeek); | |||
| } | |||
|
|
|||
| void EngineBuffer::postProcessLocalBpm() { | |||
| // This needs to be done at once for all channels | |||
There was a problem hiding this comment.
this comment feels like it should go in the place where this call happens, not in the function definition itself. (it talks about "all channels", which is the context of the call site, not the implementation.
src/engine/enginebuffer.h
Outdated
| @@ -71,6 +71,8 @@ class EngineBuffer : public EngineObject { | |||
| /// This is an artificial state that happens if a standard seek and a | |||
| /// phase seek are scheduled at the same time. | |||
| SEEK_STANDARD_PHASE = SEEK_STANDARD | SEEK_PHASE, | |||
| /// #SEEK_EXACT to the other deck position | |||
| SEEK_CLONE = 8u, | |||
There was a problem hiding this comment.
just the next enum value. I will use the bit shift notation to make this obvious.
| @@ -390,6 +390,12 @@ void EngineMixer::processChannels(int iBufferSize) { | |||
| // Syncables will not. | |||
| m_pEngineSync->onCallbackEnd(m_sampleRate, iBufferSize); | |||
|
|
|||
| for (int i = activeChannelsStartIndex; | |||
There was a problem hiding this comment.
comment:
// After all engines have been processed, trigger updates of local bpm values which may have changed based on track position
src/engine/enginemixer.cpp
Outdated
| // After all the engines have been processed, trigger post-processing | ||
| // which ensures that all channels are updating certain values at the | ||
| // same point in time. This prevents sync from failing depending on |
There was a problem hiding this comment.
| // After all the engines have been processed, trigger post-processing | |
| // which ensures that all channels are updating certain values at the | |
| // same point in time. This prevents sync from failing depending on | |
| // After local bpms are updated, trigger the rest of the post-processing | |
| // which ensures that all channels are updating certain values at the | |
| // same point in time. This prevents sync from failing depending on |
src/engine/sync/synccontrol.cpp
Outdated
| } | ||
| // This slot is fired by a new file is loaded or if the user | ||
| // has adjusted the beatgrid. | ||
| // Note: The track is loaded not not yet cued. |
There was a problem hiding this comment.
| // Note: The track is loaded not not yet cued. | |
| // Note: The track is loaded but not yet cued. |
|
Done |
| /// This is an artificial state that happens if a standard seek and a | ||
| /// phase seek are scheduled at the same time. | ||
| SEEK_STANDARD_PHASE = SEEK_STANDARD | SEEK_PHASE, | ||
| /// #SEEK_EXACT to the other deck position | ||
| SEEK_CLONE = 8u, | ||
| SEEK_CLONE = 1 << 3 |
Co-authored-by: Owen Williams <[email protected]>
|
Done |
|
@ywwg Can you merge this? |
|
@ywwg IIUC is overwhelmed by (work) github notifications, so I pinged him on Zulip. |
|
success! thanks for the direct ping I know it's an annoying extra step |
This PR fixes #10550 Via a couple of single fixes regarding using the right data at the right time.
This is a split from #12435 which also fixes gain issues happens during track cloning