Instructions to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf",
	filename="salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ3_M.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

Use Docker

docker model run hf.co/RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

LM Studio
Jan
Ollama
How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with Ollama:
```
ollama run hf.co/RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
```

Unsloth Studio

How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf to start chatting

Docker Model Runner
How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with Docker Model Runner:
```
docker model run hf.co/RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M
```

Lemonade

How to use RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf:Q4_K_M

Run and chat with the model

lemonade run user.Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf-Q4_K_M

List all available models

lemonade list

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Quantization made by Richard Erkhov.

Github

Discord

Request more models

salt-asr_wav-uni_1_tts_wav-uni_1-12k - GGUF

Model creator: https://huggingface.co/Vikhrmodels/
Original model: https://huggingface.co/Vikhrmodels/salt-asr_wav-uni_1_tts_wav-uni_1-12k/

Name	Quant method	Size
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q2_K.gguf	Q2_K	1.28GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ3_XS.gguf	IQ3_XS	1.39GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ3_S.gguf	IQ3_S	1.45GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q3_K_S.gguf	Q3_K_S	1.45GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ3_M.gguf	IQ3_M	1.5GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q3_K.gguf	Q3_K	1.58GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q3_K_M.gguf	Q3_K_M	1.58GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q3_K_L.gguf	Q3_K_L	1.7GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ4_XS.gguf	IQ4_XS	1.72GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q4_0.gguf	Q4_0	1.8GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.IQ4_NL.gguf	IQ4_NL	1.8GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q4_K_S.gguf	Q4_K_S	1.81GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q4_K.gguf	Q4_K	1.89GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q4_K_M.gguf	Q4_K_M	1.89GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q4_1.gguf	Q4_1	1.96GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q5_0.gguf	Q5_0	2.12GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q5_K_S.gguf	Q5_K_S	2.12GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q5_K.gguf	Q5_K	2.17GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q5_K_M.gguf	Q5_K_M	2.17GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q5_1.gguf	Q5_1	2.29GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q6_K.gguf	Q6_K	2.47GB
salt-asr_wav-uni_1_tts_wav-uni_1-12k.Q8_0.gguf	Q8_0	3.2GB

Original model description:

English Version 🇬🇧

Model Performance Overview

Metrics:

PESQ@200: Perceptual Evaluation of Speech Quality (higher = better).
STOI@200: Short-Time Objective Intelligibility (closer to 1 = better).
SI-SDR@200: Scale-Invariant Signal-to-Distortion Ratio (higher = better).
SIM-O@200: Similarity to ground truth (higher = better).

Model	PESQ@200	STOI@200	SI-SDR@200	SIM-O@200
Original (LibriSpeech)	4.15	0.997	27.45 ±1.09	—
Parler TTS Mini v1	1.29 ±0.49	0.15 ±0.12	25.0 ±2.9	0.88 ±0.03
Fish Speech 1.5	1.26 ±0.38	0.17 ±0.12	25.0 ±3.2	0.91 ±0.02
Salt-ASR Wav-Uni 1-12k	1.27 ±0.40	0.18 ±0.09	20.3 ±3.69	0.88 ±0.02

Our Solution

Method: Extends a pre-trained LLM with audio tokens and fine-tunes on TTS and ASR tasks.
Training:
- SpeechTokenizer (semantic + audio tokens) outperformed Encodec (loss explosions resolved with TF32 precision).
- Training time: 150 A100 GPU hours.
Advantages: Unified LM loss for dual tasks, minimal training overhead.

Resources

Code: GitHub Repo
Inference Demo: Google Colab
Reference Papers: Vitta, Valle

Русская Версия 🇷🇺

Сравнение моделей

Метрики:

PESQ@200: Качество речи (чем выше, тем лучше).
STOI@200: Разборчивость речи (ближе к 1 = лучше).
SI-SDR@200: Соотношение сигнал-шум (выше = лучше).
SIM-O@200: Сходство с эталоном (выше = лучше).

Модель	PESQ@200	STOI@200	SI-SDR@200	SIM-O@200
Original (LibriSpeech)	4.15	0.997	27.45 ±1.09	—
Parler TTS Mini v1	1.25 ±0.49	0.15 ±0.12	25.0 ±2.9	0.88 ±0.03
Fish Speech 1.5	1.26 ±0.38	0.17 ±0.12	25.0 ±3.2	0.91 ±0.02
Salt-ASR Wav-Uni 1-12k	1.27 ±0.40	0.18 ±0.09	20.3 ±3.69	0.88 ±0.02

Наше решение

Метод: Расширение словаря LLM аудиотокенами + дообучение на TTS и ASR.
Обучение:
- SpeechTokenizer (семитические + аудиотокены) показал лучшие результаты, чем Encodec.
- Время обучения: 150 часов на A100.
Преимущества: Единая функция потерь для двух задач, минимальные затраты.

Ресурсы

Код: GitHub
Демо: Google Colab

Примечание: Модель поддерживает генерацию коротких фраз на английском, немецком и французском.

Downloads last month: 501

GGUF

Model size

3B params

Architecture

llama

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for RichardErkhov/Vikhrmodels_-_salt-asr_wav-uni_1_tts_wav-uni_1-12k-gguf

VITA: Towards Open-Source Interactive Omni Multimodal LLM

Paper • 2408.05211 • Published Aug 9, 2024 • 50