docs: add cross-project usage section
Browse files
SETUP.md
CHANGED
|
@@ -67,19 +67,19 @@ Model: .../onnx/tiny-swara-int8.onnx
|
|
| 67 |
|
| 68 |
Play `hello.wav`. Use `--voice BM` for the male voice.
|
| 69 |
|
| 70 |
-
| Voice | Who
|
| 71 |
-
|---|---|
|
| 72 |
-
| `BF`
|
| 73 |
-
| `BM`
|
| 74 |
|
| 75 |
Other useful options:
|
| 76 |
|
| 77 |
-
| Option
|
| 78 |
-
|---|---|
|
| 79 |
-
| `--text-file input.txt`
|
| 80 |
-
| `--speed 1.2`
|
| 81 |
| `--model onnx/tiny-swara.onnx` | use the FP32 model instead of INT8 |
|
| 82 |
-
| `--output out.wav`
|
| 83 |
|
| 84 |
## 5. Use it from Python
|
| 85 |
|
|
@@ -90,26 +90,60 @@ tts = TinySwara(voice="BF")
|
|
| 90 |
tts.save("আমি তোমাকে পছন্দ করি।", "hello.wav")
|
| 91 |
```
|
| 92 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
## Troubleshooting
|
| 94 |
|
| 95 |
-
| Problem
|
| 96 |
-
|---|---|
|
| 97 |
-
| `espeak-ng is not installed`
|
| 98 |
-
| `ModuleNotFoundError: No module named 'tiny_voca'` | Run `python inference.py` from inside the `tiny-swara` folder
|
| 99 |
-
| `Could not find an ONNX model`
|
| 100 |
-
| Bengali text is garbled in the terminal
|
| 101 |
-
| `python3 -m venv` fails
|
| 102 |
|
| 103 |
`inference.py` loads `inventory.json`, `lexicon.json` and `tiny_voca/text.py`
|
| 104 |
relative to itself, so always run it from the repository folder.
|
| 105 |
|
| 106 |
## What's in the folder
|
| 107 |
|
| 108 |
-
| Path
|
| 109 |
-
|---|---|
|
| 110 |
-
| `onnx/tiny-swara-int8.onnx`
|
| 111 |
-
| `onnx/tiny-swara.onnx`
|
| 112 |
-
| `inference.py`
|
| 113 |
-
| `tiny_voca/text.py`
|
| 114 |
| `config.json`, `inventory.json`, `lexicon.json` | vocabulary and pronunciation data |
|
| 115 |
-
| `samples/`
|
|
|
|
| 67 |
|
| 68 |
Play `hello.wav`. Use `--voice BM` for the male voice.
|
| 69 |
|
| 70 |
+
| Voice | Who |
|
| 71 |
+
| ----- | ---------------- |
|
| 72 |
+
| `BF` | female (default) |
|
| 73 |
+
| `BM` | male |
|
| 74 |
|
| 75 |
Other useful options:
|
| 76 |
|
| 77 |
+
| Option | Meaning |
|
| 78 |
+
| ------------------------------ | ---------------------------------- |
|
| 79 |
+
| `--text-file input.txt` | read UTF-8 text from a file |
|
| 80 |
+
| `--speed 1.2` | faster speech |
|
| 81 |
| `--model onnx/tiny-swara.onnx` | use the FP32 model instead of INT8 |
|
| 82 |
+
| `--output out.wav` | where to save |
|
| 83 |
|
| 84 |
## 5. Use it from Python
|
| 85 |
|
|
|
|
| 90 |
tts.save("আমি তোমাকে পছন্দ করি।", "hello.wav")
|
| 91 |
```
|
| 92 |
|
| 93 |
+
## 6. Use it in another project
|
| 94 |
+
|
| 95 |
+
You do not have to `cd` into this folder to use the Python API. `inference.py`
|
| 96 |
+
finds `config.json`, `inventory.json`, `lexicon.json`, `tiny_voca/` and `onnx/`
|
| 97 |
+
relative to itself (not your current directory), so the folder can live
|
| 98 |
+
anywhere as long as you keep that layout intact.
|
| 99 |
+
|
| 100 |
+
Three things must exist in the other project's environment:
|
| 101 |
+
|
| 102 |
+
1. **espeak-ng** — the same system package from step 1, installed once per machine.
|
| 103 |
+
2. **Python dependencies** — in that project's virtual environment:
|
| 104 |
+
`pip install numpy onnxruntime`
|
| 105 |
+
3. **This folder** — copy or clone it somewhere stable. You can drop `samples/`
|
| 106 |
+
and the `LICENSE`/`.md` files; keep `inference.py`, `tiny_voca/`,
|
| 107 |
+
`config.json`, `inventory.json`, `lexicon.json` and `onnx/`.
|
| 108 |
+
|
| 109 |
+
Then add the folder (the one containing `inference.py`) to `sys.path` and import:
|
| 110 |
+
|
| 111 |
+
```python
|
| 112 |
+
import sys
|
| 113 |
+
sys.path.insert(0, "/path/to/tiny-swara-hf")
|
| 114 |
+
from inference import TinySwara
|
| 115 |
+
|
| 116 |
+
tts = TinySwara(voice="BF") # loads the INT8 model by default
|
| 117 |
+
tts.save("আমি তোমাকে পছন্দ করি।", "out/hello.wav")
|
| 118 |
+
```
|
| 119 |
+
|
| 120 |
+
Relative output paths resolve against your current working directory, so the
|
| 121 |
+
WAV lands wherever your project expects it.
|
| 122 |
+
|
| 123 |
+
> If you vendor the folder _inside_ your project as a subpackage, keep it on
|
| 124 |
+
> `sys.path` as above. `inference.py` imports `tiny_voca` as a top-level module,
|
| 125 |
+
> so nesting it under a package name breaks that import.
|
| 126 |
+
|
| 127 |
## Troubleshooting
|
| 128 |
|
| 129 |
+
| Problem | Fix |
|
| 130 |
+
| -------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- |
|
| 131 |
+
| `espeak-ng is not installed` | Install it (step 1) and make sure `espeak-ng --version` works in the same terminal |
|
| 132 |
+
| `ModuleNotFoundError: No module named 'tiny_voca'` | Run `python inference.py` from inside the `tiny-swara` folder |
|
| 133 |
+
| `Could not find an ONNX model` | Keep the folder layout intact; do not move `inference.py` away from `onnx/`, `config.json`, `inventory.json`, `lexicon.json` and `tiny_voca/` |
|
| 134 |
+
| Bengali text is garbled in the terminal | Save the sentence to a UTF-8 file and use `--text-file input.txt` |
|
| 135 |
+
| `python3 -m venv` fails | `sudo apt install python3-venv` |
|
| 136 |
|
| 137 |
`inference.py` loads `inventory.json`, `lexicon.json` and `tiny_voca/text.py`
|
| 138 |
relative to itself, so always run it from the repository folder.
|
| 139 |
|
| 140 |
## What's in the folder
|
| 141 |
|
| 142 |
+
| Path | Purpose |
|
| 143 |
+
| ----------------------------------------------- | --------------------------------- |
|
| 144 |
+
| `onnx/tiny-swara-int8.onnx` | recommended model, 8.24 MiB |
|
| 145 |
+
| `onnx/tiny-swara.onnx` | FP32 reference, 20.4 MiB |
|
| 146 |
+
| `inference.py` | CLI and Python API |
|
| 147 |
+
| `tiny_voca/text.py` | Bengali text-to-phoneme front end |
|
| 148 |
| `config.json`, `inventory.json`, `lexicon.json` | vocabulary and pronunciation data |
|
| 149 |
+
| `samples/` | example audio from the model card |
|