PiperTTS-DE / README.md
qaihm-bot's picture
v0.58.0
0fc0303 verified
|
Raw
History Blame Contribute Delete
15.2 kB
---
library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: text-to-audio
---
![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/web-assets/model_demo.png)
# PiperTTS-DE: Optimized for Qualcomm Devices
PiperTTS is a high-quality multi-lingual text-to-speech library.
This is based on the implementation of PiperTTS-DE found [here](https://github.com/rhasspy/piper).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/pipertts_de) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
## Getting Started
There are two ways to deploy this model on your device:
### Option 1: Download Pre-Exported Models
Below are pre-exported model assets ready for deployment.
| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| VOICE_AI | float | Snapdragon® X2 Elite | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_x2_elite.zip)
| VOICE_AI | float | Snapdragon® X Elite | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_x_elite.zip)
| VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_8gen3.zip)
| VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_8gen1.zip)
| VOICE_AI | float | Qualcomm® QCS8275 | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_qcs8275.zip)
| VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_qcs8550_proxy.zip)
| VOICE_AI | float | Qualcomm® SA8775P | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_sa8775p.zip)
| VOICE_AI | float | Snapdragon® 8 Elite Mobile | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_8_elite_for_galaxy.zip)
| VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_snapdragon_8_elite_gen5_for_galaxy.zip)
| VOICE_AI | float | Qualcomm® SA7255P | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_sa7255p.zip)
| VOICE_AI | float | Qualcomm® SA8295P | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_sa8295p.zip)
| VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/pipertts_de/releases/v0.58.0/pipertts_de-voice_ai-float-qualcomm_qcs9075.zip)
For more device-specific assets and performance metrics, visit **[PiperTTS-DE on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/pipertts_de)**.
### Option 2: Export with Custom Configurations
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/pipertts_de) Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations
This option is ideal if you need to customize the model beyond the default configuration provided here.
See our repository for [PiperTTS-DE on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/pipertts_de) for usage instructions.
## Model Details
**Model Type:** Model_use_case.audio_generation
**Model Stats:**
- Model checkpoint: rhasspy/piper-checkpoints
- Max decoded sequence length: 64 tokens
- Number of parameters (encoder): 7.51M
- Model size (encoder) (float): 28.7 MB
- Number of parameters (sdp): 1.04K
- Model size (sdp) (float): 6.61 KB
- Number of parameters (flow): 7.39M
- Model size (flow) (float): 28.2 MB
- Number of parameters (decoder): 1.66M
- Model size (decoder) (float): 6.37 MB
- Number of parameters (t5_encoder): 15.1M
- Model size (t5_encoder) (float): 57.5 MB
- Number of parameters (t5_decoder): 5.72M
- Model size (t5_decoder) (float): 21.8 MB
## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| charsiu_decoder | VOICE_AI | float | Snapdragon® X2 Elite | 0.334 ms | 1 - 1 MB | NPU
| charsiu_decoder | VOICE_AI | float | Snapdragon® X Elite | 0.415 ms | 1 - 1 MB | NPU
| charsiu_decoder | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 0.309 ms | 0 - 7 MB | NPU
| charsiu_decoder | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 0.572 ms | 1 - 13 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® QCS8275 | 0.975 ms | 0 - 8 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.398 ms | 0 - 2 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® SA8775P | 0.647 ms | 0 - 9 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® SA8650P | 0.647 ms | 0 - 9 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® SA8255P | 0.647 ms | 0 - 9 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® QCS8450 | 0.572 ms | 1 - 13 MB | NPU
| charsiu_decoder | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.26 ms | 0 - 9 MB | NPU
| charsiu_decoder | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 0.272 ms | 0 - 13 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® SA7255P | 0.975 ms | 0 - 8 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 0.546 ms | 3 - 5 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® SA8295P | 0.795 ms | 0 - 6 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 0.272 ms | 0 - 13 MB | NPU
| charsiu_decoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 0.415 ms | 1 - 1 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® X2 Elite | 0.624 ms | 0 - 0 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® X Elite | 1.064 ms | 0 - 0 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 0.632 ms | 0 - 7 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 1.353 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® QCS8275 | 2.75 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.886 ms | 0 - 1 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® SA8775P | 1.264 ms | 0 - 10 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® SA8650P | 1.264 ms | 0 - 10 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® SA8255P | 1.264 ms | 0 - 10 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® QCS8450 | 1.353 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.488 ms | 0 - 8 MB | NPU
| charsiu_encoder | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 0.529 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® SA7255P | 2.75 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 1.116 ms | 0 - 2 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® SA8295P | 1.752 ms | 0 - 6 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 0.529 ms | 0 - 9 MB | NPU
| charsiu_encoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 1.064 ms | 0 - 0 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® X2 Elite | 1.778 ms | 0 - 0 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® X Elite | 3.088 ms | 0 - 0 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 2.189 ms | 0 - 7 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 4.387 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® QCS8275 | 8.141 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 3.011 ms | 0 - 2 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® SA8775P | 3.461 ms | 0 - 10 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® SA8650P | 3.461 ms | 0 - 10 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® SA8255P | 3.461 ms | 0 - 10 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® QCS8450 | 4.387 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.822 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 1.898 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® SA7255P | 8.141 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 3.332 ms | 0 - 2 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® SA8295P | 3.989 ms | 0 - 6 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 1.898 ms | 0 - 9 MB | NPU
| decoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 3.088 ms | 0 - 0 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® X2 Elite | 18.671 ms | 0 - 0 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® X Elite | 29.322 ms | 0 - 0 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 23.738 ms | 0 - 7 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 38.348 ms | 1 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® QCS8275 | 47.396 ms | 0 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 31.494 ms | 0 - 2 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® SA8775P | 32.991 ms | 0 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® SA8650P | 32.991 ms | 0 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® SA8255P | 32.991 ms | 0 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® QCS8450 | 38.348 ms | 1 - 10 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 17.098 ms | 0 - 8 MB | NPU
| encoder | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 19.324 ms | 0 - 9 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® SA7255P | 47.396 ms | 0 - 10 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 33.026 ms | 0 - 3 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® SA8295P | 35.24 ms | 0 - 6 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 19.324 ms | 0 - 9 MB | NPU
| encoder | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 29.322 ms | 0 - 0 MB | NPU
| flow | VOICE_AI | float | Snapdragon® X2 Elite | 9.69 ms | 4 - 4 MB | NPU
| flow | VOICE_AI | float | Snapdragon® X Elite | 15.87 ms | 4 - 4 MB | NPU
| flow | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 10.982 ms | 4 - 11 MB | NPU
| flow | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 18.195 ms | 3 - 13 MB | NPU
| flow | VOICE_AI | float | Qualcomm® QCS8275 | 40.096 ms | 1 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 15.057 ms | 4 - 7 MB | NPU
| flow | VOICE_AI | float | Qualcomm® SA8775P | 17.152 ms | 1 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® SA8650P | 17.152 ms | 1 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® SA8255P | 17.152 ms | 1 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® QCS8450 | 18.195 ms | 3 - 13 MB | NPU
| flow | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 8.91 ms | 1 - 9 MB | NPU
| flow | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 8.925 ms | 3 - 16 MB | NPU
| flow | VOICE_AI | float | Qualcomm® SA7255P | 40.096 ms | 1 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 17.75 ms | 4 - 10 MB | NPU
| flow | VOICE_AI | float | Qualcomm® SA8295P | 18.953 ms | 0 - 6 MB | NPU
| flow | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 8.925 ms | 3 - 16 MB | NPU
| flow | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 15.87 ms | 4 - 4 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® X2 Elite | 7.132 ms | 0 - 0 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® X Elite | 11.293 ms | 0 - 0 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® 8 Gen 3 Mobile | 7.689 ms | 0 - 8 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® 8 Gen 1 Mobile | 11.166 ms | 0 - 10 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® QCS8275 | 20.589 ms | 0 - 10 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 10.408 ms | 0 - 2 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® SA8775P | 11.144 ms | 3 - 9 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® SA8650P | 11.144 ms | 3 - 9 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® SA8255P | 11.144 ms | 3 - 9 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® QCS8450 | 11.166 ms | 0 - 10 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.603 ms | 0 - 9 MB | NPU
| sdp | VOICE_AI | float | Snapdragon® 8 Elite Mobile | 6.95 ms | 0 - 13 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® SA7255P | 20.589 ms | 0 - 10 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-9075 | 11.075 ms | 2 - 4 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® SA8295P | 12.664 ms | 0 - 6 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® Dragonwing™ Q-8750 | 6.95 ms | 0 - 13 MB | NPU
| sdp | VOICE_AI | float | Qualcomm® Dragonwing™ IQ-X7181 | 11.293 ms | 0 - 0 MB | NPU
## License
* The license for the original implementation of PiperTTS-DE can be found
[here](https://github.com/rhasspy/piper/blob/master/LICENSE.md).
## References
* [PiperTTS High-quality Multi-lingual Multi-accent Text-to-Speech](https://github.com/rhasspy/piper)
* [Source Model Implementation](https://github.com/rhasspy/piper)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).