F5-TTS-English-ONNX

F5-TTS English — ONNX export

A ready-to-use ONNX Runtime export of the base F5-TTS (F5TTS_v1_Base) checkpoint, packaged for Horus.F5Tts.Onnx — run F5-TTS text-to-speech from .NET with no Python. Exported at NFE 32 (the library default).

Files

• F5_Preprocess.onnx, F5_Transformer.onnx, F5_Decode.onnx
• vocab.txt

Usage (.NET)

using var model = F5TtsModel.Load(
    "F5_Preprocess.onnx", "F5_Transformer.onnx", "F5_Decode.onnx", "vocab.txt");

var (reference, _) = WavAudio.ReadPcm16("reference_en.wav"); // 24 kHz mono 16-bit PCM
var result = model.Synthesize(reference,
    referenceText: "The transcript of the reference clip.",
    text: "Hello, how are you today?");
File.WriteAllBytes("out.wav", result.ToWav());

English works with the library's default character-level tokenizer. See the library README for GPU providers and options.

Credits & license

• Model: F5-TTS (F5TTS_v1_Base) by SWivid.
• ONNX export tooling: DakeQQ/F5-TTS-ONNX.
• ONNX files re-hosted from H5N1AIDS/F5-TTS-ONNX, with the matching F5TTS_v1_Base vocab.txt added.
• License: CC-BY-NC-4.0 — the base F5-TTS weights are non-commercial (SWivid/F5-TTS).