michi-ai 0.1.0__py3-none-any.whl → 0.1.1__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (5) hide show
  1. {michi_ai-0.1.0.dist-info → michi_ai-0.1.1.dist-info}/METADATA +4 -4
  2. michi_ai-0.1.1.dist-info/RECORD +6 -0
  3. michi_ai-0.1.0.dist-info/RECORD +0 -6
  4. {michi_ai-0.1.0.dist-info → michi_ai-0.1.1.dist-info}/LICENSE.txt +0 -0
  5. {michi_ai-0.1.0.dist-info → michi_ai-0.1.1.dist-info}/WHEEL +0 -0
{michi_ai-0.1.0.dist-info → michi_ai-0.1.1.dist-info}/METADATA RENAMED
@@ -1,12 +1,12 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: michi-ai
3
- Version: 0.1.0
3
+ Version: 0.1.1
4
4
  Summary: Full-duplex speech LLM client for MichiAI
5
5
  Home-page: https://ketsuilabs.io
6
6
  License: Apache-2.0
7
7
  Keywords: llm,speech,full-duplex,audio
8
8
  Author: Damian Krystkiewicz
9
- Author-email: damian@ketsuilabs.io
9
+ Author-email: 45499236+dkrystki@users.noreply.github.com
10
10
  Requires-Python: >=3.9,<4.0
11
11
  Classifier: License :: OSI Approved :: Apache Software License
12
12
  Classifier: Programming Language :: Python :: 3
@@ -40,7 +40,7 @@ Unlike traditional serial pipelines (ASR → LLM → TTS), MichiAI can listen an
40
40
  | **Latency (TTFA)** | ~75ms (tested on RTX 4090) |
41
41
  | **Architecture** | Continuous Embeddings + Rectified Flow Matching |
42
42
  | **Base Backbone** | SmolLM-360m |
43
- | **Key Innovation** | No Coherence Loss / Single Forward Pass per Word |
43
+ | **Key Innovation** | No Coherence Loss / Single Step Decoding |
44
44
 
45
45
 
46
46
  ## 🌟 Key Features
@@ -56,7 +56,7 @@ Unlike traditional serial pipelines (ASR → LLM → TTS), MichiAI can listen an
56
56
  ## 🤖 Architecture Overview
57
57
 
58
58
  ### 1. The Listening Head
59
- A multi-modal encoder mapping raw audio into a continuous embeddings while simultaneously generating text tokens. This ensures the model understands both the semantic meaning and the emotional context.
59
+ A multi-modal encoder mapping raw audio into continuous embeddings while simultaneously generating text tokens. This ensures the model understands both the semantic meaning and the emotional context.
60
60
 
61
61
  ### 2. The Speaking Head
62
62
  Predicts audio embeddings using **Rectified Flow Matching**. This allows for fast, high-quality, and diverse speech generation. The embeddings are then processed through a lightweight, causal **HiFi-GAN vocoder** for real-time streaming.
michi_ai-0.1.1.dist-info/RECORD ADDED
@@ -0,0 +1,6 @@
1
+ michiai/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
2
+ michiai/client.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
3
+ michi_ai-0.1.1.dist-info/LICENSE.txt,sha256=TGMMvGlYQvvAXSjlrZvpXJjiaxnwDsarhDmLY1KIWr0,11351
4
+ michi_ai-0.1.1.dist-info/METADATA,sha256=Kxrbe0rKJ4EEiiaJLPHrtxYuJQuOZPFqEa8yBC9odjU,4133
5
+ michi_ai-0.1.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
6
+ michi_ai-0.1.1.dist-info/RECORD,,
michi_ai-0.1.0.dist-info/RECORD DELETED
@@ -1,6 +0,0 @@
1
- michiai/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
2
- michiai/client.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
3
- michi_ai-0.1.0.dist-info/LICENSE.txt,sha256=TGMMvGlYQvvAXSjlrZvpXJjiaxnwDsarhDmLY1KIWr0,11351
4
- michi_ai-0.1.0.dist-info/METADATA,sha256=oKqjq5PuboN4yquaUsRi2pX-IsmUXjy2WRtCRUpDo80,4121
5
- michi_ai-0.1.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
6
- michi_ai-0.1.0.dist-info/RECORD,,