TADA: Fast, Reliable Speech Generation Through Text-Acoustic Synchronization(www.hume.ai)
40 pointsby smusamashah4 hours ago5 comments
  • qinqiang201an hour ago
    Could it run on Macbook? Just on GPU device?
  • OutOfHere2 hours ago
    Will this run on CPU? (as opposed to GPU)
    • boxedan hour ago
      Why would you want to? It's like using a hammer for screws.
      • g-morkan hour ago
        CPU compute is infinity times less expensive and much easier to work with in general
        • boxed36 minutes ago
          Less expensive how? The reason GPUs are used is because they are more efficient. You CAN run matmul on CPUs for sure, but it's going to be much slower and take a ton more electricity. So to claim it's "less expensive" is weird.
      • regularfryan hour ago
        To maximise the VRAM available for an LLM on the same machine. That's why I asked myself the same question, anyway.
  • octoclaw5 minutes ago
    [dead]
  • theturtle35 minutes ago
    [dead]
  • zacklee19883 hours ago
    [dead]