Dev gzf exp (#1707)

zhifu gao

2024-05-08 4adb76a6edbca93aae7caa83382e764d7b058f07

 funasr/models/sense_voice/model.py

@@ -516,16 +516,23 @@

        # Paramterts for rich decoding
        self.beam_search.emo_unk = tokenizer.encode(
            DecodingOptions.get("emo_unk_token", "<|SPECIAL_TOKEN_1|>"), allowed_special="all")[0]
            DecodingOptions.get("emo_unk_token", "<|SPECIAL_TOKEN_1|>"), allowed_special="all"
        )[0]
        self.beam_search.emo_unk_score = 1
        self.beam_search.emo_tokens = tokenizer.encode(
            DecodingOptions.get("emo_target_tokens", "<|HAPPY|><|SAD|><|ANGRY|>"), allowed_special="all")
            DecodingOptions.get("emo_target_tokens", "<|HAPPY|><|SAD|><|ANGRY|>"),
            allowed_special="all",
        )
        self.beam_search.emo_scores = DecodingOptions.get("emo_target_threshold", [0.1, 0.1, 0.1])

        self.beam_search.event_bg_token = tokenizer.encode(
            DecodingOptions.get("gain_tokens_bg", "<|Speech|><|BGM|><|Applause|><|Laughter|>"), allowed_special="all")
            DecodingOptions.get("gain_tokens_bg", "<|Speech|><|BGM|><|Applause|><|Laughter|>"),
            allowed_special="all",
        )
        self.beam_search.event_ed_token = tokenizer.encode(
            DecodingOptions.get("gain_tokens_ed", "<|/Speech|><|/BGM|><|/Applause|><|/Laughter|>"), allowed_special="all")
            DecodingOptions.get("gain_tokens_ed", "<|/Speech|><|/BGM|><|/Applause|><|/Laughter|>"),
            allowed_special="all",
        )
        self.beam_search.event_score_ga = DecodingOptions.get("gain_tokens_score", [1, 1, 1, 1])

        encoder_out, encoder_out_lens = self.encode(
@@ -859,16 +866,23 @@

        # Paramterts for rich decoding
        self.beam_search.emo_unk = tokenizer.encode(
            DecodingOptions.get("emo_unk_token", "<|SPECIAL_TOKEN_1|>"), allowed_special="all")[0]
            DecodingOptions.get("emo_unk_token", "<|SPECIAL_TOKEN_1|>"), allowed_special="all"
        )[0]
        self.beam_search.emo_unk_score = 1
        self.beam_search.emo_tokens = tokenizer.encode(
            DecodingOptions.get("emo_target_tokens", "<|HAPPY|><|SAD|><|ANGRY|>"), allowed_special="all")
            DecodingOptions.get("emo_target_tokens", "<|HAPPY|><|SAD|><|ANGRY|>"),
            allowed_special="all",
        )
        self.beam_search.emo_scores = DecodingOptions.get("emo_target_threshold", [0.1, 0.1, 0.1])

        self.beam_search.event_bg_token = tokenizer.encode(
            DecodingOptions.get("gain_tokens_bg", "<|Speech|><|BGM|><|Applause|><|Laughter|>"), allowed_special="all")
            DecodingOptions.get("gain_tokens_bg", "<|Speech|><|BGM|><|Applause|><|Laughter|>"),
            allowed_special="all",
        )
        self.beam_search.event_ed_token = tokenizer.encode(
            DecodingOptions.get("gain_tokens_ed", "<|/Speech|><|/BGM|><|/Applause|><|/Laughter|>"), allowed_special="all")
            DecodingOptions.get("gain_tokens_ed", "<|/Speech|><|/BGM|><|/Applause|><|/Laughter|>"),
            allowed_special="all",
        )
        self.beam_search.event_score_ga = DecodingOptions.get("gain_tokens_score", [1, 1, 1, 1])

        encoder_out, encoder_out_lens = self.encode(