alexmrqt / gr-lazyviterbi Goto Github PK

Hello,

I'm interested in running an HMM on long (~100,000) sequences to decode ~10 different states. I can use a normal Viterbi algorithm to decode this. However, i'm interested in speeding up the processing and would like to try a lazy viterbi algorithm. This is one of the few open source modules i have found and would like to try this out. However, i'm completely unfamiliar with GNU Radio. I have tried playing around with the examples and i'm a bit stymied. I have a few questions i was hoping you could answer.

1. Could this be used to run a HMM to decode this data? My understanding is that GNU Radio is normally used on streaming data, however, in your example you seem to show how to create a stream from vector data.

2. Is it possible to specify the state model in your lazy viterbi class to contain multiple states with custom transition probabilities?

3. How do you grab the decoded data? Currently, in one of the examples, you use a null sink. Is the data stored there somehow? Do i need to use a different sink to grab the data to process? Any advice on how to do this?

THanks for any help you can give me.